PROTOCOL TITLE:

Response:

Application of Natural Language Processing to Improve Care for Aortic Aneurysm Patients at
Penn State Hershey Medical Center

PROTOCOL VERSION/AMENDMENT # AND DATE

Response:

Initial submission

PRINCIPAL INVESTIGATOR:
Response:

TBD

CO-INVESTIGATORS:
Response:

TBD

DATE:
Response:

1.0 Objectives

1.1 Describe the purpose, specific aims, or objectives of this research. Specifically,
explain why it is important to do the study.

Response:

Page 1 of 39
PROTOCOL TITLE:

Purpose: : Improve detection of AA in the Stony Brook Medicine patient population by using
NLP-based free text analysis of the electronic medical record (EMR) and a regularly updated AA
data mart to facilitate patient surveillance, promote risk factor modification strategies, optimize
delivery of timely care, and prevent AA-related emergencies.

* Of note, the term “aortic aneurysms (AAs)” will be used rather than “abdominal aortic aneurysm (AAAs)” to
denote the fact that our study will include both thoracic and abdominal aortic aneurysms.

Importance of Study:
Current screening strategies for AAs are solely based on patient age and smoking
history, with patient follow-up plans determined only by the size of aneurysms at
the time of detection. As a result, large patient groups (ie. females, patients < 65
years old) are excluded from screening, and individual risk of AA progression has
not been clearly defined. Despite the abundance of medical data captured and
stored in EMR systems, the ability to meaningfully analyze this data to refine risk
profiles and advance screening practices has remained elusive. Fortunately,
improvement in AI capabilities has made it possible to apply machine learning
algorithms to massive sets of medical data.
Our work combines discrete EHR data elements, natural language processing
(NLP) of computed tomography (CT) reports and computer vision applied to CT
images, creating “multi-modal” data pipelines. To predict AA growth, we have
developed methods in computer vision, NLP, and fluid structure interaction (FSI)
models to conduct integrative statistical analyses. This collaborative effort
combines clinical, biomedical informatics (BMI), biomedical engineering (BME)
and statistical expertise to develop novel algorithms to better manage AA.
In this study, we aim to improve the identification and care of asymptomatic AA patients
through the application of validated machine learning algorithms to EMR-captured data.
The implications of successful implementation are profound. The ability to leverage AI
for the extraction, processing, and analyzing of big data would not only help elucidate
unexplored patterns and associations, but also help deliver a personalized approach to
screening and intervention based on the patient’s individual risk factors. Successful
implementation of AI would allow for earlier identification of high risk for aggressive
disease progression, early risk factor modification, and timely elective intervention to
mitigate risk of rupture.

1.2 State the hypothesis to be tested, if applicable.

NOTE: A hypothesis is a specific, testable prediction about what you expect to
happen in your study that corresponds with your above listed objectives.

Page 2 of 39
PROTOCOL TITLE:

Response:

Primary Hypothesis:
We hypothesize that integrative analysis of multi-modal data collected through the use of natural
language processing (NLP), computer vision models (CV), and patient-specific FSI models
developed by our group, will allow us to leverage the combined clinical, imaging data, and
biomechanical data for predicting AA growth and treatment outcome.

AI-assisted chart review can also improve identification, monitoring, and care of patients with
aneurysmal disease by ensuring timely surveillance and intervention.

Secondary Hypothesis:
Application of both algorithms will demonstrate significant reduction in time required to
complete chart review tasks relative to manual chart review, the implications of which are
tremendous cost savings for the health care system.

2.0 Scientific/Safety Endpoints

2.1 Describe the scientific endpoint(s), the main result or occurrence under study.
NOTE: Scientific endpoints are outcomes defined before the study begins to determine
whether the objectives of the study have been met and to draw conclusions from the data.
Include primary and secondary endpoints. Some example endpoints are: reduction of
symptoms, improvement in quality of life, or survival. Your response should not be a
date.
Response:

Scientific endpoints in this study will objectively measure the development and validation of
machine learning algorithms:

- Endpoint for NLP algorithm

o Successful detection of AA presence within free text reports in the EMR and
accurate extraction of discrete data (i.e. maximal aneurysm diameter, morphologic
characteristics)
- End point for CV algorithm
o Accuracy of aortic segmentation (as assessed via calculation of DICE score)
o Successful identification of imaging biomarkers in CT scan
o Extraction of accurate geometric features of AA (i.e. tortuosity and undulation
measurements)

Page 3 of 39
PROTOCOL TITLE:

o Successful prediction of patients with higher likelihood of aneurysmal disease

progression via analysis of CT Scan

3.0 Background

3.1 Provide the scientific or scholarly background, rationale, and significance of the
research based on the existing literature and how it will contribute/fill in gaps to existing
knowledge.

Response:

Abdominal aortic aneurysms (AAA) are a pathologic dilatation of the abdominal aorta beyond
50% of the vessel’s diameter, generally 3.0 cm or larger. While most AAAs are completely
asymptomatic and diagnosed incidentally, if left untreated, progressive aneurysm growth can
lead to rupture, a catastrophic complication with mortality of up to 80%. Of those with a
ruptured AAA, only 50% reach a hospital, and of those, only 50% survive emergent repair [1,3].
Comparatively, when identified early in the disease process and before rupture occurs, elective
aneurysm repair carries a postoperative mortality of 1.2-3.2% [7]. Given this disparity, there
have been strong efforts to promote vascular screening programs. These efforts paired with
widespread public and preventative health initiatives were intended to promote earlier detection
and treatment for AAA - the results of which amount to a >50% decrease in the incidence of
death over the last 20 years [6].

Current screening guidelines, established by the Society of Vascular Surgery recommend a one-
time ultrasound screening for AAAs in men or women 65 to 75 years of age with a history of
tobacco use [6]. The timing of follow-up surveillance studies is determined by measured
aneurysm size. For aneurysms measured 3-3.4 cm repeat imaging is performed in 3 years, 3.5-
4.4 cm yearly, and 4.5-5.4 cm every 6 months [4]. For men with an aneurysm measuring >5.5 cm
and women >5.0 cm, surgical intervention is advised [6].

However, current surveillance recommendations are rudimentary, solely basing screening

strategies on patient age and smoking history. Current guidelines exclude large patient groups
such as females and younger patients, and have no consideration of individual patient-level risk
factors not yet explored; individual risk of AA progression has not been clearly defined. The
benefit of current screening programs remain unclear, as evidence suggest that current
preventative strategies have failed to reduce the incidence of aortic rupture. The pathophysiology
of AAA has been extensively studied and described, which include oxidative stress, immune-
mediated factors, genetic factors, and others. However, it is generally accepted that the causes
are multifactorial and the specific mechanisms of aneurysmal disease and progression have yet to

Page 4 of 39
PROTOCOL TITLE:

be elucidated. Without the discovery of new risk factors or imaging biomarkers, vascular
surgeons have found it difficult to improve on current preventative strategies for patients with
aneurysmal disease. This is particularly true for patients with ectasia or mild aneurysmal disease,
for whom interval imaging can be as long as 3 years.

Fortunately, recent advances in machine learning algorithms, increases in computational speed

with GPUs (Graphical Processing Units), and development of specialized software packages
allow for levels of data analysis and pattern recognition previously inaccessible to researchers.
With respect to the field of vascular surgery, recent work by Hirata et al. describes using
machine learning to explore the use of various computed tomography (CT) angiography
measurements of aortic aneurysms to predict aneurysm expansion [11], with similar work
completed by Akkoyun et al. [12]. Lu et al. used deep learning methods to segment the aorta and
identify AAAs from the radiographic CT images with better sensitivity than radiologists [13].
Garcia-Garcia et al. explored the use of clinical elements and image-derived features to classify
patients as having ‘slow', 'medium', or "quick' aneurysm growth rates [9].

We will expand upon the work performed on radiology image analysis and aneurysm
progression analysis to develop a risk progression model based on direct analysis of radiologic
images and clinical data elements. We will compare the performance of these advanced models
including convolutional neural networks, recurrent neural networks to more traditional machine
learning techniques, such as regularized logistic regression and random forest. In addition to
identifying patients at increased risk of aneurysm progression, by using EHR data in conjunction
with the PACS images, we will be able identify patients at increased risk of having an aneurysm.

Successful implementation of these powerful tools can not only help to improve overall care of
AA patients, but also has great academic and economic implications. AI models and algorithms,
as they continue to become more refined and capable, have already changed the way researchers
tackle big data. As adoption continues to accelerate, research studies previously deemed too
laborious, costly, and impractical to pursue will become much more feasible and less cost-
prohibitive. Given the ever-tightening NIH and healthcare budget, the value of AI as a tool for
drastically reducing the cost to conduct research cannot be overstated.

3.2 Include complete citations or references:

Response:

1. Hoornweg LL, Storm-Versloot MN, Ubbink DT, Koelemay MJ, Legemate

DA, Balm R. Meta analysis on mortality of ruptured abdominal aortic
aneurysms. Eur J Vasc Endovasc Surg 2008;35:558

Page 5 of 39
PROTOCOL TITLE:

2. Kent KC. Clinical practice. Abdominal aortic aneurysms. N Engl J Med

2014;371:2101-8.
3. Aggarwal S QA, Sharma V, Sharma A. Abdominal aortic aneurysm: A
comprehensive review. Experimental and Clinical Cardiology 2011;16:11-5.
4. LCWK1. Deaths, percent of total deaths, and death rates for the 15 leading
causes of death in 5 year age groups, by race and sex: United States, 2015.
2017:10. 5.
5. Li X, Zhao G, Zhang J, Duan Z, Xin S. Prevalence and trends of the
abdominal aortic aneurysms epidemic in general population--a meta-analysis.
PLoS One 2013;8:e81260. 6.
6. Chaikof EL, Dalman RL, Eskandari MK, et al. The Society for Vascular
Surgery practice guidelines on the care of patients with an abdominal aortic
aneurysm. Journal of Vascular Surgery 2018;67:2-77.e2. 7.
7. Qadura M, Pervaiz F, Harlock JA, et al. Mortality and reintervention
following elective abdominal aortic aneurysm repair. J Vasc Surg
2013;57:1676-83, 83 el. 8.
8. Kent KC, Zwolak RM, Egorova NN, et al. Analysis of risk factors for
abdominal aortic aneurysm in a cohort of more than 3 million individuals.
Journal of Vascular Surgery 2010;52:539-48. 9.
9. Garcia-Garcia F ME, Christodoulidis S, Anthimopoulos M, Kontopodis N,
Correa-Londono M, Wyss TR, Papaharilaou Y, Ioannou CV, Tengg-Kobligk
HV, Mougiakakou SG. Prognosis of Abdominal Aortic Aneurysms: A
Machine Learning-Enabled Approach Merging Clinical, Morphometric,
Biomechanical and Texture Information. 2017 IEEE 30th International
Symposium on Computer-Based Medical Systems (CBMS) 2017:463-8.
10. Creager MA, Loscalzo J. Diseases of the Aorta. In: Jameson JL, Fauci AS,
Kasper DL, Hauser SL, Longo DL, Loscalzo J, eds. Harrison's Principles of
Internal Medicine, 20e. New York, NY: McGraw-Hill Education; 2018. 11.
11. Hirata K, Nakaura T, Nakagawa M, et al. Machine Learning to Predict the
Rapid Growth of Small Abdominal Aortic Aneurysm. Journal of Computer
Assisted Tomography 2020:44:37-42. 12.
12. Akkoyun E, Kwon ST, Acar AC, Lee W, Baek S. Predicting abdominal aortic
aneurysm growth using patient-oriented growth models with two-step
Bayesian inference. Computers in Biology and Medicine 2020;117:103620.

Page 6 of 39
PROTOCOL TITLE:

13. Lu J-T, Brooks R, Hahn S, et al. DeepAAA: Clinically Applicable and

Generalizable Detection of Abdominal Aortic Aneurysm Using Deep
Learning. Springer International Publishing; 2019:723-31.

4.0 Study Design

4.1 Describe and explain the study design (e.g. case-control, cross-sectional,
ethnographic, experimental, interventional, longitudinal, and observational). Indicate if
there is randomization, blinding, control group, etc. If randomizing, explain how this will
be achieved.

Response:

A retrospective and ongoing chart review of patients with AA (abdominal or thoracic aortic
aneurysm), seen in vascular surgery clinic at Stony Brook Medicine from January 1, 2010 to
December 31, 2018, will be conducted. To do this we will query the Stony Brook EMR for all
patients with diagnosis of AA, seen during this time interval.

This retrospective cohort study will consist of 3 groups:

1. Regular patients followed in clinic for period of time (denoted by multiple clinic
visits), and without vascular intervention (open or endovascular) due to AAA
diameter not meeting threshold for intervention

2. Patients followed in clinic with substantial progression of aneurysmal disease,

meeting size or growth rate criteria for intervention.

3. Patients who came to clinic to with aneurysm size meeting size of repair.

5.0 Local Number of Subjects

5.1 Indicate the total number of subjects who will be enrolled or records that will be
reviewed through Stony Brook.

Response:

Query yielded a patient list consisting of 1,954 patients seen in clinic from January 1, 2010 to
December 31, 2019, 1018 patients (52.1%) had multiple clinic visits.

Page 7 of 39
PROTOCOL TITLE:

5.2 If this study is only being conducted through Stony Brook, provide statistical
justification (i.e. power analysis) for the number of subjects provided in 5.1 above. If
qualitative research, so state, and provide general justification for the total number of
subjects proposed.
Response:
NA. No reliable pilot data or evidence to establish standard deviation.

5.3 If applicable, indicate your screen failure rate, i.e., how many subjects you expect
to screen to reach your target sample.
Response:
NA. No reliable pilot data or evidence to establish standard deviation.

5.4 Justify the feasibility of recruiting the proposed number of eligible subjects within
the anticipated recruitment period. For example, how many potential subjects do you
have access to? What percentage of those potential subjects do you need to recruit?
Response:

Response:
NA. No reliable pilot data or evidence to establish standard deviation.

6.0 Inclusion and Exclusion Criteria

NOTE: If your study is more than minimal risk, you must also upload a copy of your
inclusion/exclusion checklist (with space for specific subject values) to be completed at
time of enrollment of each subject.

6.1 Describe, in bullet points, the criteria that define who will be included in this
study:

Response:

Inclusion Criteria:

Page 8 of 39
PROTOCOL TITLE:

Vascular surgery patients with:

- Diagnosis of ectasia or aneurysm of the abdominal or thoracic aorta as indicated by ICD
code
- Diagnosis of ectasia or aneurysm of the abdominal or thoracic aorta as indicated within
the free text of clinical notes, radiology reports, or scanned notes imported into EMR
- Imaging studies with abdominal aorta measuring >2.5 cm or ascending thoracic aorta
measuring >4.0 cm or descending thoracic aorta measuring >3.0 cm
*** Inclusion criteria subject to change as study progresses (ie. Inclusion of patients with other imaging studies such
as magnetic resonance imaging (MRI) or ultrasound (US))

6.2 Describe, in bullet points, the criteria that define who will be excluded from this
study:

Response:

Exclusion Criteria:
Vascular surgery patients with:
- Diagnosis of AA via ICD code but no imaging or mention of aneurysmal disease within
clinical documentation
- Imaging studies with unique factors likely to lower the quality and accuracy of CV model
training
o Poor quality studies
 ie. imaging artifact, concurrent abdominal pathology, incomplete study (ie.
Aneurysm cut off inferiorly on chest CT)

6.3 Describe how individuals will be screened for eligibility. Upload all relevant
screening documents with your submission (screening protocol, script, questionnaire).
Identify who will certify that subjects meet eligibility requirements.

Response:

N/A – This is a records review. No patient screening or recruitment procedures.

6.4 Indicate whether you are specifically recruiting or targeting any of the following
special populations in your study using the checkboxes below. (You will be asked for
additional information in Section 7 if you check any of these boxes)

Response:

Page 9 of 39
PROTOCOL TITLE:

N/A – This is a records review. No patient screening or recruitment procedures.

Adults unable to consent

Minors (under 18 years old)
Pregnant women
Prisoners

6.5 Indicate if you will include minorities (American Indians, Alaskan Native, Asian,
Native Hawaiian, Pacific Islander, Black [not of Hispanic origin] and Hispanic) as
Federal mandates require that you include minorities unless you can justify their
exclusion
Response:

✘ Yes
No, Justify:

6.6 Indicate whether you will include non-English speaking individuals in your study.
Provide justification if you will specifically exclude non-English speaking individuals.
Review http://research.stonybrook.edu/human-subjects-standard-operating-procedures/
policy-non-english-speakers-research-subjects for SBU policy on inclusion of non-
English speakers. Upload any translated materials (consent, questionnaires, etc).

Response:

N/A – This is a records review. No patient screening or recruitment procedures.

7.0 Vulnerable Populations

7.1 For research that involves pregnant women, review, complete and upload
Supplemental Form A: Pregnant Women, Fetuses, Non-Viable Neonates , or Neonates of
Uncertain Viability.
Confirmed
✘ N/A: This research does not involve pregnant women.

7.2 For research that involves neonates of uncertain viability or non-viable neonates,
review, complete and upload Supplemental Form A: Pregnant Women, Fetuses, Non-
Viable Neonates, or Neonates of Uncertain Viability.
Confirmed

Page 10 of 39
PROTOCOL TITLE:

✘ N/A: This research does not involve non-viable neonates or neonates of uncertain
viability.

7.3 For research that involves prisoners, review, complete and upload Supplemental
Form H: Prisoners
Confirmed
✘ N/A: This research does not involve prisoners.

7.4 For research that involves minors (under 18 years), review, complete and upload
Supplemental Form F: Minors
Confirmed
✘ N/A: This research does not involve persons who have not attained the legal age for
consent to treatments or procedures (“children”).

7.5 For research that involves adults who cannot consent for themselves, you will be
asked additional information in Section 25 (“Informed Consent”)
Confirmed
✘ N/A: This research does not involve this population

7.6 Consider if other specifically targeted populations such as students, employees of

a specific firm, or educationally or economically disadvantaged persons are vulnerable.
Provide information regarding their safeguards and protections, including safeguards to
eliminate coercion or undue influence.
Safeguards include:

✘ N/A

8.0 Eligibility Screening

8.1 Describe screening procedures for determining subjects’ eligibility. Screening

refers to determining if prospective participants meet inclusion and exclusion criteria.
Include all relevant screening documents with your submission (e.g. screening
protocol, script, questionnaires) as attachments.
Response:

✘ N/A: There is no screening as part of this protocol.

9.0 Recruitment Methods

✘ N/A: This is a records review only, and subjects will not be recruited. NOTE: If you
select this option, please make sure that all records review procedures and inclusion/exclusion

Page 11 of 39
PROTOCOL TITLE:

screening are adequately described in other sections, including date range for records that will be
reviewed.

9.1 Describe source of subjects: When, where, and how potential subjects will be
recruited.
NOTE: Recruitment refers to how you are identifying potential participants and
introducing them to the study. These may include, but are not limited to:
ResearchMatch.org, physician referral, Office of Clinical Trials database, West Campus
departmental pools, reviewing medical charts, Research Participant Groups/help
groups, advertising companies, call centers, in person announcements / presentations
Response:

N/A – This is a records review. No patient screening or recruitment procedures.

9.2 Describe how you will protect the privacy interests of prospective subjects during
the recruitment process.
NOTE: Privacy refers to an individual’s right to control access to him or herself. This is
NOT asking about confidentiality of data.

Response:

N/A – This is a records review. No patient screening or recruitment procedures.

9.3 Identify/describe any materials that will be used to screen/recruit subjects and
upload copies of these documents with the application. They may include, but are not
limited to Telephone scripts for calling, flyers, Questionnaires, Posters, Letters or
written material to be sent or emailed, pamphlets, posted advertisements, email
invitations.

Response:

N/A – This is a records review. No patient screening or recruitment procedures.

10.0 Research Procedures

Provide a detailed description of all research procedures or activities being performed on the
research subjects. This should serve as a blueprint for your study and include enough detail
so that another investigator could pick up your protocol and replicate the research. For

Page 12 of 39
PROTOCOL TITLE:

studies that have multiple or complex visits or procedures, consider the addition of a schedule of
events table in in your response. Be sure to include:
 Procedures being performed to monitor subjects for safety or to minimize risks.
 All drugs and devices used in the research and the purpose of their use, and their
regulatory status

Response:

Research Procedure:

Phase 1 – Query, Data Extraction,

In the first phase of this study, we will conduct a chart review of vascular surgery clinic
encounters from January 1st 2010 to present. To obtain patient list for review, a query of the
Stony Brook EMR will filter for patients seen in this time interval and also with diagnosis of
abdominal aortic aneurysms, thoracic abdominal aortic aneurysms, and aortic ectasia.

Data extraction for this patient list will include medical, demographic, and radiologic data
including but not limited to currently well-known risk factors for AA, other comorbidities,
surgeries, procedures, family history, social history, radiologic studies, laboratory values, and
medications. Customized query of HealtheIntent database will be coded by team members from
bioinformatics department.

This list will be made available to the research team and grouped based on inclusion criteria for
each phase of this study. Extensive manual chart review will be conducted in order to confirm
accurate query results, establish “ground truth”, and organize data into usable form for training
algorithms.

*** Inclusion criteria may be modified as study progresses and pathophysiologic associations are uncovered.

Phase 2 – Identification

Natural Language Processing (NLP) Algorithm Testing and Validation

- NLP algorithms will be used to identify patients with AA (binary data), and determine
size of aneurysm, specifically diameter (continuous data). To do this, the algorithm will
scan and analyze free-text notes in the EMR, specifically radiology reports.
- Successful identification of AA diagnosis and accurate measurement of aneurysm
diameter will be determined by comparing algorithm results to manual chart review
(ground truth data).
- Accuracy, sensitivity, and specificity of algorithm will be measured.

Page 13 of 39
PROTOCOL TITLE:

*** It should be noted there was no direct patient interaction in association with this study, even
in the case of incidental findings of AA. In accordance with standard practice, proper due
diligence will be carried out and patient’s PCP will be notified via mail informing them of the
abnormal findings. ***

Computer Vision (CV) Algorithm Testing and Validation

Investigation of clinical data alone may not be adequate for elucidating all contributing factors
for AA disease progression. Imaging and simulation-based assessments of the aorta suggest
certain geometric or biomechanical markers are predictive of aggressive AA growth. However,
the majority of studies investigating computer-vision techniques are limited by small sample
sizes and inadequate longitudinal data for small aneurysms. We have leveraged our data
collection methods to address these shortcomings and conduct thorough, sufficiently powered
investigations.

Steps for computer vision model testing and validation will include the following:
- Filter will be applied to queried patient list to identify patients who have undergone
multiple CT scans, or a CT scan followed by an abdominal ultrasound (US) study
- 3D medical imaging (i.e. CT scans) will be de-identified and aorta will be “annotated” to
create manual aortic segmentations. These segmentations serve as training data for the
computer vision model to learn how to “see” the aorta as humans do, and distinguish it
from the rest of the surrounding anatomy
- Annotated ground truth data segmentation will be reviewed by vascular clinicians.
Aneurysms will be measured manually in initial and subsequent scans. Interval growth
will be noted and converted to Δmm/yr.
- “High risk” patients will be determined by interval increase of > 3mm/yr or progression
to size threshold for intervention (5-5.5cm)

Computer Vision Assessment of AA Morphology and Geometry:

- The model will also include additional features including morphologic assessment and
geometric feature extraction. Automatic image analysis will allow for center-line based
accurate measurements along the length of the aorta. This is important as it allows for the
creation of customized geometric indices that help quantify the degree of tortuosity,
undulation, and surface irregularity of diseased portions.
- Additional CV features include assessing presence of calcification, presence of thrombus,
and other possible pathologic biomarkers.
- After developing and validating CV models for the accurate identification of each
imaging biomarker, machine learning algorithms can be used to evaluate biomarkers with
higher correlation with rapid aneurysm growth in subsequent studies

Page 14 of 39
PROTOCOL TITLE:

- CV methods that will be explored includes the use of model for semantic segmentation of
the aorta followed by precise geometric feature extraction and the use of Vision
Transformers (ViT), a more recent computer vision method that may be trained to make
growth predictions based on the raw imaging data, thus obviating the need for aortic
segmentation.
- Performance of these CV models will be assessed by determining the agreement and
degree of error while predicting aneurysmal growth.

** (CV Analysis Security Protocol: CT scans in the form of DICOM files will be coded before
analysis. Analysis of imaging will be conducted behind HIPAA firewall).

- CV implementation will include the identification of unique morphologic features of the

aneurysm such as geometry, degree of tapering, presence of calcification, presence of
thrombus, etc.
o Proper training of CV models to accurately identify each unique imaging
biomarker, will require multiple steps of image segmentation

Geometric, Hemodynamic, and Biomechanical Assessment of AA

- Aortic segmentations produced by CV models facilitate the application of more advanced
technologies. This includes complex geometric feature extraction and patient-specific
computational flow dynamics (CFD) / fluid-structure interaction (FSI) models, which are
used for deriving the hemodynamic and biomechanical forces involved in the AA
development.
- To characterize hemodynamic/biomechanical parameters associated with AA
development (such as wall shear stress, Von Mises stress, tensile strain, and oscillatory
shear index), FSI models will be developed based on patient-specific geometry and
echocardiogram-derived flow data (i.e. peak aortic valve velocity).
o Two domains will be created: a fluid domain to represent the blood volume
(incompressible Newtonian fluid under laminar flow) and a solid domain
representing the arterial wall (linear elastic material). The outer surface of the
fluid domain and the inner surface of the solid domain will be defined as the
fluid-structure interface.
o The two domains will be fully coupled, using an arbitrary LangrangianEulerian
method. Wall shear stress, shear stress derivatives, and pressure distribution will
be derived from the fluid domain. Von Mises stress and tensile strains will be
derived from the solid domain.
- Aneurysmal growth will be assessed by comparing models created after patient follow-
up. Derived wall shear stress and wall stress/strain will be evaluated for relationships

Page 15 of 39
PROTOCOL TITLE:

with growth rate. Combined with morphologic and geometric indices, the parameters
derived from the FSI models will be used to predict AA growth.
- Using hemodynamics to predict AA development has shown promising results, and FSI
assessment can provide additional information about the wall integrity and compliance
changes.

Phase 3 – Advanced Statistical Methods for AA Growth Prediction Establishing Risk

Profiles
- After collection of all query, NLP, CV, and FSI model-derived results, we will conduct
integrative analysis of this multi-modal data
- Traditional statistical methods will be used to begin preliminary analysis and rudimentary
prediction. This will include conducting multivariate analysis to identify risk factors that
significantly correlate with rapid growth rate and uncover new patterns and associations.
Multiple linear regression can be used to make growth predictions.
- In addition to using these traditional statistical methods, machine learning methods that
can combine imaging and discrete clinical data elements will be explored. We shall
develop a novel AA-size indexed recurrent neural network as illustrated in Figure 6. Here
Xi is the input vector for each patient at the given measured AA size of i (mm), fi is the
transfer function at the given AA size, and Yi is the AA growth rate measured or
predicted at the given AA size i (mm). This framework will overcome the difficulty of a
time-indexed neural network because for our study, and nearly all longitudinal
biomedical studies, we have irregularly spaced time series measurements which will
render the classical neural network moot. The AA growth size can be easily computed
from the AA growth rate.

Future Studies
- Future studies will involve:
o Development of additional features for both NLP and CV models. This includes:
 NLP: improved localization of anatomy when extracting measurements
and morphologic features (i.e. suprarenal vs infrarenal)
 CV: novel geometric indices that focus on specific portions of the aorta
that are of interest for evaluating risk of endograft failure (i.e. proximal
and distal landing zones)
o The study of an additional cohort of patients: Post-operative patients and patients
who have come to clinic meeting the size threshold for elective surgical
intervention.
 Similar to previous studies, medical and radiologic data will be parsed and
evaluated using NLP and CV algorithms.
 But in addition to this, data will include post-operative complications,
secondary procedures, and post-operative disease progression.

Page 16 of 39
PROTOCOL TITLE:

** During the data extraction and AI-model training phase of this study, selected records will be
screened by reviewers with clinical background when evaluating progression of disease over
time. The targeted chart review will be done to assess efficacy and accuracy of analytic
techniques developed. This effort will evaluate AI methodology as a tool of accurately predicting
progression of aneurysmal disease.

All data will be stored and reviewed behind the SBM firewall. All code used to analyze the data
will be stored in the SBM GitHub repository, or other approved repositories behind SBM
firewall.

Computer Vision (CV) Analysis Security Protocol - CT scans in the form of DICOM files will be
coded before analysis. Analysis of imaging will be conducted behind HIPAA firewall **

*** It should be noted there was no direct patient interaction in association with this study, even
in the case of incidental findings of AA. In accordance with standard practice, proper due
diligence will be carried out and patient’s PCP will be notified via mail informing them of the
abnormal findings. ***

Table 1. Risk factors to be extracted from EMR via customized query coded by Bioinformatics
department

Data to be Extracted
Demographics
- Age
- Gender
- Race

Co-morbidities
- HTN, HLD, CAD, COPD
- Other aneurysmal disease
- Other vascular disease
- DVT, PE, hypercoag, etc.

Family Hx
- AAA, TAAA, dissection

Page 17 of 39
PROTOCOL TITLE:

- Other aneurysmal disease

- Other vascular disease
- Auto-immune disease
- Cancer, etc

Social Hx
- Smoking
- (Current, former,
social/intermittent, never)
- Pack-years
- Years since quitting
- Smoking interval in years

Procedures:
- Previous vascular surgery
- Coronary stents or CABG
- Other surgeries

Medications
Imaging Imaging Biomarkers
- CT, MRI , US --- (NLP) - Aneurysm diameter
- CT --- (Computer Vision) - Location (distance from
renals, bifurcation)
- Morphologic features
(inflammation, wall thickness,
tapering, etc)
-

10.1 Describe what data, including long-term follow-up, will be collected.

NOTE: For studies with multiple data collection points or long-term follow up, consider
the addition of a schedule or table in your response.

Response:

Page 18 of 39
PROTOCOL TITLE:

- Data will include risk factors for aneurysmal disease as well as other medical and
demographic information that may be potential risk factors not yet known
o Demographics
o Family Hx
o Social history
o Co-morbidities
o Surgical history (ie previous abdominal surgery)
o Labs
o Medications
o Diagnostic imaging studies

- In later phases, data from long-term follow up may include patients who have not yet had
a subsequent imaging study to assess aneurysm growth

List, and upload, any instruments or measurement tools used to collect data (e.g. survey,
scripts, questionnaire, interview guide, validated instrument, data collection form).

Response:

Data collection will include manual chart review and machine learning algorithms, specifically
natural language processing and computer vision algorithms

10.2 Describe any source records that will be used to collect data about subjects (e.g.
school records, electronic medical records) and include the date range for records that
will be accessed.

Response:

The primary source of data will be the EMR of Stony Brook Medicine with date range of eligible
records spanning the inception of the EMR to the current date. As the study has an important
quality improvement component in its detection of AA patients who may have otherwise been
missed, this study has the potential to continue collecting data into perpetuity.

Collection of data will include querying the EMR for patients with an ICD code for AA disease
as well as any clinical documentation with free text indicating the presence of AA. This would
cover patients who do indeed have AA disease but for some reason are not properly labeled with
the appropriate ICD code.

Page 19 of 39
PROTOCOL TITLE:

For the purposes of evaluating this cohort against a control group, chart review will also include
patients with risk factors for AA, but do not have AA.

** As study progresses, patient visits 2019-present may be included **

10.3 Indicate whether or not the results for individual subjects, such as results of
investigational diagnostic tests, genetic tests, or incidental findings will be shared with
subjects or others (e.g., the subject’s primary care physician) and if so, describe how
these will be shared.

Response:

Study results will not be shared with patients. In case of incidental abnormal findings, patient’s
PCP to be notified via mail.

10.4 Indicate whether or not generalized study results will be shared with subjects or
others, and if so, describe how these will be shared.

Response:

Generalized study results will not be shared with patients.

11.0 Study Timelines

11.1 Describe the anticipated duration of the study needed to enroll all study subjects.

Response:

There is no subject enrollment. This retrospective-prospective study is open-ended until this

protocol is terminated.

11.2 Describe the duration of an individual subject’s participation in the study.

Include length of study visits, and overall study follow-up time.

Response:

Page 20 of 39
PROTOCOL TITLE:

There is no subject enrollment. This retrospective-prospective study is open-ended until this

protocol is terminated.

11.3 Describe the estimated duration for the investigators to complete this study (i.e.
all data is collected and all analyses have been completed).
Response:

There is no subject enrollment. This retrospective-prospective study is open-ended until this

protocol is terminated.

12.0 Research Setting

12.1 Describe all facilities/sites/locations where you will be screening and conducting
research procedures. Include a description of the security and privacy of the facilities
(e.g. locked facility, limited access, privacy barriers). Facility, department, and type of
room are relevant. Do not abbreviate facility names.
Example: “A classroom setting in the Department of Psychology equipped with a
computer with relevant survey administration software,” “The angiogram suite at
Stony Brook University Hospital, a fully accredited tertiary care institution within
New York State with badge access,”

Response:

All data will be collected, stored, and analyzed at Stony Brook University via password-
protected computers and servers within the hospital network.

Data management will initially be conducted via use of excel sheets shared over the protected
SBU Box accounts. In time, the database will need to scale with the influx of new patients and
clinical/imaging information. For this we will set up a REDCAP account to utilize their
relational database, and data entry tools.

CT scans in the form of DICOM files will be coded before analysis. Analysis of imaging will be
conducted behind HIPAA firewall)

12.2 For research procedures being conducted, for this study, external to SBU and its
affiliates (e.g., in schools, out-of-state, internationally, etc.) describe:
 Site-specific regulations or customs affecting the research

Page 21 of 39
PROTOCOL TITLE:

 The composition and involvement of any community advisory board

 Local scientific and ethical review structure outside the organization.
 Local issues affecting the research and rights of research subjects.
NOTE: This question is not referring to multi-center research. If this research is
being conducted internationally, Supplemental Form C must be completed and
uploaded.
Response:

✘ N/A: This study is not conducted outside of SBU or its affiliates.

13.0 Resources and Qualifications

13.1 The Principal Investigator (PI) must confirm, in consultation with Chair and
Dean as applicable, that adequate resources are present to conduct and complete the
study compliantly and safely. Specifically:
NO ✘ YES The proposed subject population(s) are available in sufficient numbers to
meet the study requirements
NO ✘ YES Sufficient funds are available to conduct and complete the study
compliantly and safely
NO ✘ YES The PI and study team have sufficient time to conduct and complete the
study compliantly and safely
NO ✘ YES The PI has determined that the named study team is qualified to conduct
the research compliantly and to monitor the safety and welfare of the enrolled research
subjects effectively.
NO ✘ YES The PI ensures that the study team is fully aware of his/her involvement
in this study and the details of the study protocol
NO ✘ YES The PI ensures that the study teams will only be involved in research
procedures for which they have been trained, and are currently certified and/or licensed,
if required..

13.2 Describe the availability of medical or psychological resources that subjects

might need as a result of anticipated consequences of the human research, if applicable.
(e.g, “on-call availability of a counselor or psychologist for a study that screens subjects
for depression”).

Response:

N/A – This is a records review.

Page 22 of 39
PROTOCOL TITLE:

13.3 Describe your process to ensure that all study team members are updated on the
progress of the research and the regulatory requirements (including enrolled subjects,
unanticipated problems etc.)

Response:

Regular team meetings to be held to monitor progress, refine protocols, troubleshoot, plan next
steps, and ensure consistent progress under direction of PI

- Interdisciplinary team meetings between vascular surgery department, bioinformatics

department, and computer science department to be held biweekly
- Department team meetings to be held biweekly
- Junior researcher meetings to be held ad hoc (daily vs weekly) depending on tasks and
phase of project

14.0 Other Approvals

14. List approvals that will be obtained prior to commencing the research (e.g.,
University Hospital sign-offs per the UH Application, Cancer Center Scientific review,
school, external site, funding agency, laboratory, Radiation Safety, IBC, SCRO, IACUC,
RDRC).

Response:

University Hospital sign-offs per the UH application

N/A: This study does not require any other approvals.

15.0 Provisions to Protect the Privacy Interests of Subjects

15.1 Describe how you will protect subjects’ privacy interests during the course of this
research and any steps you will take to make the subject feel at ease.
NOTE: Privacy refers to an individual’s desire/right to control access to or to place
limits on whom they interact with or whom they provide personal information. Privacy
applies to the person. Confidentiality refers to how data collected about individuals for
the research will be protected by the researcher from release. Confidentiality applies to
the data.
Examples of appropriate responses include: “participant only meets with a study
coordinator in a private office setting where no one can overhear”, or “the participant is
reminded that they are free to refuse to answer any questions that they do not feel
comfortable answering.”

Page 23 of 39
PROTOCOL TITLE:

Response:

N/A - This is a records review.

(Confidentiality will be ensured via review of data on encrypted devices and correspondence
between research members over end-to-end encryption standards that satisfy HIPAA guidelines).

16.0 Data Management and Analysis

16.1 Describe the data analysis plan, including any statistical procedures. This
section applies to both quantitative and qualitative analysis.

Response:

Validation of NLP and CV Models will be measured by accuracy (percent error), sensitivity, and
specificity. These values will be obtained by comparing models to manual chart review, which
will serve as the ground truth data set.

Multivariate analysis of AI algorithm results as well as medical and radiologic data will be
performed to evaluate for patterns and strong correlations. This pilot study will determine the
viability of AI-assisted chart review as a valuable tool for data acquisition and analysis, but also
a valuable prediction tool for identifying high risk patients and determining the likelihood of
disease progression.

16.2 If applicable, provide a power analysis.

NOTE: This may not apply to certain types of studies, including chart/records reviews,
survey studies, or observational studies. This question is asked to elicit whether the
investigator has an adequate sample size to achieve the study objectives and justify a
conclusion.

Response:

Power analysis not applicable to retrospective portion of study, but may be applicable to future
prospective studies. However, no reliable pilot data or evidence to establish standard deviation.

17.0 Confidentiality

Page 24 of 39
PROTOCOL TITLE:

A. Confidentiality/Security of Study Data

Describe the local procedures for maintenance of security and confidentiality of study
data and any records that will be reviewed for data collection.

17.1 Where and how will all data and records be stored? Include information about:
password protection, encryption, physical controls, authorization of access, certificates
of confidentiality, and separation of identifiers and data, as applicable. Include physical
(e.g. paper) and electronic files.

Response:

All data and code generated for analysis will be kept behind the SBM firewall. Access will be
limited to those on the study.

All review of data will be on encrypted devices. Correspondence between research members will
be carried out over end-to-end encryption standards that satisfy HIPAA guidelines

17.2 How long will the data be stored?

Response:

5 years after results are published.

17.3 Who will have access to the data?

Response:

Study members

17.4 Who is responsible for receipt or transmission of the data?

Response:

Study members

17.5 How will the data be transported/transmitted?

Page 25 of 39
PROTOCOL TITLE:

Response:

Transmission of all data will be behind the SBM firewall. No data with PHI will leave SBM.

B. Confidentiality of Study Specimens

Describe the local procedures for maintenance of confidentiality of study specimens.

✘ N/A: No specimens will be collected or analyzed in this research.

(Skip to Section 18.0)

17.6 Where and how will all specimens be stored? Include information about:
physical controls, authorization of access, and labeling of specimens, as applicable.
Response:

17.7 How long will the specimens be stored?

Response:

17.8 Who will have access to the specimens?

Response:

17.9 Who is responsible for receipt or transmission of the specimens?

Response:

17.10 How will the specimens be transported?

Response:

18.0 Provisions to Monitor the Data to Ensure the Safety of Subjects

✘ N/A: This study is not enrolling subjects OR is limited to records review procedures
only OR is a minimal risk study

18.1 Describe the plan to evaluate the data periodically regarding both harms and
benefits to determine whether subjects remain safe. The plan might include establishing a

Page 26 of 39
PROTOCOL TITLE:

data safety monitoring committee and a plan for reporting data monitoring committee
findings to the IRB and the sponsor.
Response:

18.2 Describe what data are reviewed, including safety data, untoward events, and
efficacy data.
Response:

18.3 Describe any primary or secondary safety endpoints.

Response:

18.4 Describe how the safety information will be collected (e.g., with case report
forms, at study visits, by telephone calls with participants).
Response:

18.5 Describe the frequency of safety data collection, including when safety data
collection starts.
Response:

18.6 Describe who will review the safety data.

Response:

18.7 Describe the frequency or periodicity of review of cumulative safety data.

Response:

18.8 Describe the statistical tests for analyzing the safety data to determine whether
harm is occurring.
Response:

18.9 Describe any conditions that trigger an immediate suspension of the research.
Response:

Page 27 of 39
PROTOCOL TITLE:

19.0 Withdrawal of Subjects

✘ N/A: This study is not enrolling subjects. This section does not apply.

19.1 Describe anticipated circumstances under which subjects may be withdrawn from
the research without their consent.
Response:

19.2 Describe any procedures for orderly termination.

NOTE: Examples may include return of study drug, exit interview with clinician.
Include whether additional follow up is recommended for safety reasons for physical or
emotional health.
Response:

19.3 Describe procedures that will be followed when subjects withdraw from the
research, including retention of already collected data, and partial withdrawal from
procedures with continued data collection, as applicable.
Response:

19.4 Describe what will happen to data already collected.

Response:

20.0 Risks to Subjects

20.1 In your opinion, what is the overall risk (physical and nonphysical) to research
subjects in this study (minimal, greater than minimal or unknown)

Response:

N/A

20.2 Describe if any subjects are withdrawn from therapeutic procedures or drugs (e.g.,
washout periods) prior to, or during, their participation in the study.

Response:

N/A

Page 28 of 39
PROTOCOL TITLE:

20.3 List the reasonably foreseeable risks, discomforts, hazards, or inconveniences to the
subjects related to their participation in the research. Consider physical, psychological,
social, legal, and economic risks. Include a description of the probability, magnitude,
duration, and reversibility of the risks.
NOTE: Breach of confidentiality is always a risk for identifiable subject data.

Response:

Breach of confidentiality
- Probability low. All precautions stated above will be implemented to ensure no breach of
confidentiality.

20.4 Describe procedures performed to minimize the probability or magnitude of risks,

including procedures being performed to monitor subjects for safety.

Response:

Data will be stored and analyzed within hospital network with password protection. All data will
be coded when possible to minimize contact with PHI.

CV Analysis Security Protocol - CT scans in the form of DICOM files will be coded before
analysis. Analysis of imaging will be conducted behind HIPAA firewall.

20.5 If applicable, indicate which procedures may have risks to the subjects that are
currently unforeseeable.

Response:

N/A

20.6 Indicate which research procedures, if any, may have risks to an embryo or fetus
should the subject be or become pregnant.

Response:

Page 29 of 39
PROTOCOL TITLE:

N/A

✘ N/A

20.7 If you responded to 20.6 that there are such risks, how will you minimize the risk of
a pregnancy occurring during the course of the study? (Select all that apply)
Counseling on birth control and /or abstinence
Pregnancy test during the study
Pregnancy test prior to initiation of the study
Other _____
✘ N/A

20.8 If applicable, describe possible risks to others who are not subjects.

Response:

N/A

21.0 Potential Benefits to Subjects

21.1 Describe the potential benefits that individual subjects may experience by taking
part in the research. Include the probability, magnitude, and duration of the potential
benefits.

Response:

N/A

21.2 Indicate if there is no direct benefit.

NOTE: Compensation cannot be stated as a benefit.

Response:

N/A

Indicate if there is a potential benefit to others, future science or society.

Response:

Page 30 of 39
PROTOCOL TITLE:

N/A

22.0 Compensation for Research-Related Injury

✘ N/A: The research procedures for this study do not present risk of research related
injury. This section does not apply.

22.1 If the research procedures carry a risk of research related injury, describe the
available compensation to subjects in the event that such injury should occur.

Response:

N/A

22.2 Provide a copy of contract language, if any, relevant to compensation for

research related injury.
NOTE: If the contract is not yet approved at the time of this submission, submit the
current version here. If the contract is later approved with different language regarding
research related injury, you must modify your response here and submit an amendment
to the IRB for review and approval.
Response:

N/A

23.0 Economic Burden to Subjects

23.1 Describe any costs that subjects may be responsible for because of participation
in the research.
NOTE: Some examples include transportation or parking.

Response:

N/A

Page 31 of 39
PROTOCOL TITLE:

✘ N/A: This study is not enrolling subjects, or is limited to records review procedures only.
This section does not apply.

24.0 Compensation for Participation

✘ N/A: There is no compensation for participation. This section does not apply.

24.1 Describe the amount/nature and timing/scheduling of any compensation to

subjects, including monetary, course credit, or gift card compensation. Describe any
prorated payments based on participation.

Response:

N/A

24.2 Justify the amount and scheduling of payments described above to ensure that
they are reasonable and commensurate with the expected contributions of the
participant. If multiple visits are involved payments should be prorated.
Note: If using West Campus Departmental pools, participation in studies may be offered
for credit in class but students MUST be given other options for fulfilling the research
component that are comparable in terms of time, effort, and education benefit. Please
list alternative activities

Response:

N/A

25.0 Informed Consent

25.1 Will you be obtaining consent from subjects?

Yes (If yes, Provide responses to each question in this Section, and upload
your consent documents where indicated in the electronic submission system)
✘ No (If no, Skip to next section)

25.2 Describe how the capacity to consent will be assessed for all subjects. Review for
guidance http://research.stonybrook.edu/human-subjects-standard-operating-
procedures/determining-potential-adult-subjects-ability-consent:

Response:

Page 32 of 39
PROTOCOL TITLE:

N/A

25.3 Describe the consent process that will be conducted to ensure that subject is fully
informed regarding study details and subject rights. Include where the consent process
will take place, with consideration of the need to protect the subject’s right to privacy.

Response:

N/A

25.4 Describe how you will ensure that subjects are provided with sufficient time to
consider taking part in the research study. Detail if there is there any time period
expected between informing the prospective subject and obtaining the consent.
NOTE: It is respectful to the prospective subject to ensure that sufficient time is given to
have their questions answered and to consider their participation

Response:

N/A

25.5 Describe the process to ensure ongoing consent, defined as a subject’s

willingness to continue participation for the duration of the research study.

Response:

N/A

Non-English Speaking Subjects

N/A: This study will not enroll Non-English speaking subjects.

25.6 Indicate which language(s) other than English are likely to be spoken/understood
by your prospective study population or their legally authorized representatives.

Response:

N/A

Page 33 of 39
PROTOCOL TITLE:

25.7 If subjects who do not speak English will be enrolled, describe the process to
consent the subjects, as well as the process to be used to ensure their understanding of
research procedures throughout the conduct of the study. Review SOP’s section 17.8 for
important policies in this regard: http://research.stonybrook.edu/human-subjects-
standard-operating-procedures/policy-non-english-speakers-research-subjects for SBU
policy on inclusion of non-English speakers.

Response:

N/A

Adults Unable to Consent

✘ N/A: This study will not enroll adults unable to consent.

25.8 Justify why it is necessary to include adult subjects who are unable to consent.
Response:

N/A

25.9 Describe how you will identify Legally Authorized Representatives (LAR) for the
subjects that will be consistent with the NYS Family Health Care Decisions Act
(FHCDA; see http://research.stonybrook.edu/human-subjects-standard-operating-
procedures/definitions-2). Indicate why it is necessary to include subjects who are
unable to consent.
Note: For research conducted outside of New York State, provide information that
describes which individuals are authorized under applicable law to consent on behalf of
a prospective subject to their participation in the research.
Response:

N/A

25.10 Describe the process for obtaining assent from the adult subjects
Indicate whether assent will be obtained from all, some, or none of the subjects. If some,
indicate which adults will be required to assent and which will not.

Response:
N/A

Page 34 of 39
PROTOCOL TITLE:

If assent will not be obtained from some or all subjects, provide an explanation of why
not.
Response:

N/A

25.11 Describe whether assent of the adult subjects will be documented and the process
to document assent.

Response:

N/A

25.12 Describe how you will obtain consent from a subject to use their data if they later
become capable of consent. How will competence be assessed and by whom?

Response:

N/A

26.0 Waiver or Alteration of Consent Process

Complete this section if:
 Informed consent will not be obtained at all
 Informed consent will be obtained, but not documented, or
 consent will be obtained, but not all required information will be disclosed (e.g., in
deception research)
N/A: A waiver or alteration of consent is not being requested.

26.1 Review, complete, and upload SUPPLEMENTAL FORM G: Consent Waivers

✘ Confirmed

26.2 If the research involves a waiver of the consent process for planned emergency
research, please contact the Office of Research Compliance for guidance regarding
assistance in complying with federal regulations governing this activity (see:
https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/CFRSearch.cfm?fr=50.24)

Page 35 of 39
PROTOCOL TITLE:

27.0 Multi-Site Research (Multisite/Multicenter Only)

✘ N/A: This study is not an investigator-initiated, multi-site study. This section does
not apply.

27.1 If this is a multi-site study where SBU is the lead site and/or the IRB of record,
describe the processes to ensure communication among sites. Include:
 All sites have the most current version of the IRB documents, including the
protocol, consent document, and HIPAA authorization.
 All required approvals have been obtained at each site (including approval by the
site’s IRB of record).
 All modifications have been communicated to sites, and approved (including
approval by the site’s IRB of record) before the modification is implemented.
 All engaged participating sites will safeguard data as required by local
information security policies.
 All local site investigators conduct the study appropriately.
 All non-compliance with the study protocol or applicable requirements will be
reported in accordance with local policy.

Response:

N/A

27.2 Describe the method for communicating to engaged participating sites:

 Problems
 Interim results
 Study closure

Response:
N/A

27.3 Indicate and statistically justify the total number of subjects that will be enrolled
or records that will be reviewed across all sites.

Response:

N/A

Page 36 of 39
PROTOCOL TITLE:

28.0 Banking Data or Specimens for Future Unspecified Use

✘ N/A: This study is not storing data or specimens for research outside the scope of
the present protocol. This section does not apply.

IMPORTANT: If you are proposing to bank specimens for future use, you may be
subject to licensure requirements under the NYS Department of Health, and must
be covered under the SBU license. See SOPs at
http://research.stonybrook.edu/human-subjects-standard-operating-procedures/
data-tissue-registries-banks

28.1 If data will be banked for research outside of the scope of the present protocol,
describe where the data will be stored, how long they will be stored, how will they be
accessed, and who will have access to the data
NOTE: Your response here must be consistent with the information provided to subjects
in your Consent Documents

Response:

N/A

28.2 If specimens will be banked (stored) for research outside of the scope of the
present protocol, describe where the specimens will be stored, how long they will be
stored, identifiers that will be associated with each specimen, how will they be accessed,
and who will have access to the specimens

NOTE: Your response here must be consistent with the information provided to subjects
in your
Consent Documents

Response:

N/A

28.3 Describe the procedures to release banked data and/or specimens for future uses,
including: the process to request a release, approvals required for release, who can
obtain data or specimens, and the data to be provided with specimens.

Response:
N/A

Page 37 of 39
PROTOCOL TITLE:

29.0 Drugs and Devices

✘ N/A: This study does not involve drugs or devices. This section does not apply.

29.1 Does this study involve use of radiopharmaceuticals? Yes No

29.2 For investigational devices (including marketed devices being used off label),
Provide the following information below:
Where will the device(s) be stored? Note that the storage area must be within an area
under the PI’s control
Describe the security of the storage unit/facility
Provide full detail regarding how the dispensing of the device(s) will be controlled
(accountability of removal/return of used devices, and disposition of remaining devices
at the conclusion of the investigation) and documented (accounting records/logs)
Response:

N/A

29.3 For investigational drugs (including marketed drugs being used off label), will
the services of the Investigational Drug Pharmacy be used for storage, dispensing,
accounting the drug (required for research conducted at UH, HSC, Cancer Center, and
Ambulatory Surgery Center)?

Yes
No →PI Provide the following information below:
 Where will the drugs/biologics be stored? Note that the storage area must be
within an area under the PI’s control
 Describe the security of the storage unit/facility:
 Provide full detail regarding dispensing of the drugs(s), how labeled, controlled
(accountability, disposition of unused drug at the conclusion of the investigation)
and documented (accounting records/logs):
Response:

N/A

30.0 Sharing of Results with Subjects

Page 38 of 39
PROTOCOL TITLE:

30.1 Describe whether results (study results or individual subject results, such as
results of investigational diagnostic tests, genetic tests, or incidental findings) will be
shared with subjects or others (e.g., the subject’s primary care physicians) and if so,
describe how it will be shared.
Response:

N/A

Page 39 of 39