UNIVERSITY OF ZIMBABWE
FACULTY OF COMPUTER ENGINEERING, INFORMATICS AND COMMUNICATION
DEPARTMENT OF COMPUTER SCIENCE
CAPSTONE PROJECT 1 (HCS460)
Literature Review:
Mental Health Chatbot for Emotional Support and Resource Guidance
PARTIAL FULFILLMENT OF THE DEGREE OF BACHELOR OF
SCIENCE HONORS IN COMPUTER SCIENCE
STUDENT NAME: PRAISE TAKUDZWA JAVANGWE
REG NUMBER: R216976K
PROGRAM: HCS
SUPERVISOR: MR. DEVE
ASSESSOR: MR. MURAMBA
YEAR OF GRAD: 2025
Contents
Introduction................................................................................................................................3
Theoretical Framework .............................................................................................................3
Historical Context .....................................................................................................................3
Key studies and findings ...........................................................................................................4
Gaps in research ........................................................................................................................4
Research Methodology .............................................................................................................5
The Software Development Life Cycle for Mind-mate.............................................................6
Resource Requirements ............................................................................................................7
Hardware Requirements ........................................................................................................7
Software Requirements .........................................................................................................7
Human Resources...................................................................................................................8
Functional Requirements ..........................................................................................................8
Non-functional Requirements....................................................................................................9
Project Plan .............................................................................................................................10
Important Tasks and Schedule.............................................................................................10
Gantt Chart...............................................................................................................................11
Use-case diagram: Mental Health Chatbot .............................................................................12
Conclusion ..............................................................................................................................12
References................................................................................................................................13
Introduction
Mental health is still a major worldwide issue, and conventional support networks are
frequently hampered by things like resource scarcity, stigma, and geographic restrictions.
Mental health chatbots have become scalable solutions that can offer crisis intervention
and easily available emotional support in order to close these gaps (World Health
Organization,
2020). MindMate is an AI-powered chatbot designed specifically for Zimbabwe that
combines multimodal sentiment analysis, which includes voice and facial expression
analysis, with multilingual capabilities, especially in Shona. In order to meet Zimbabwe's
particular mental health requirements, this project intends to develop a culturally appropriate
support network, thereby supporting the expanding field of digital mental health therapies.
Theoretical Framework
This project is based on the ideas of human-computer interaction (HCI), cognitive behavioral
theory (CBT), and cross-cultural psychology. Many digital mental health therapies
are supported by cognitive behavioral theory, which offers organized therapy
approaches
including reframing negative beliefs. Chatbots can provide these approaches in a
nonjudgmental and helpful manner (Wykes & Schueller, 2015). According to human-
computer interaction theory, creating sympathetic user interfaces that genuinely engage users
is crucial for increasing user satisfaction and retention. Last but not least, MindMate's
language and sentiment elements are designed with cross-cultural psychology in mind,
acknowledging that cultural influences have a significant impact on emotional expression and
mental health experiences (Prensky, 2005). Together, these frameworks direct MindMate's
strategy for offering culturally competent mental health assistance.
Historical Context
From early healthcare initiatives to AI-powered chatbots, the use of digital tools in mental
health has changed throughout time. The first attempt to use natural language processing
to simulate therapeutic talks was made in the 1960s with the introduction of ELIZA.
ELIZA's capabilities were restricted to scripted answers, but this development set the
foundation for
future mental health applications (Bickmore & Picard, 2005). Rapid advances inAI over the
last 20 years have made it possible to create complex chatbots for mental health, like Wysa
and Woebot, which combine AI-driven empathy simulations with evidence-based treatment
approaches (Chai & Lee, 2020). The necessity for culturally inclusive models like MindMate,
which supports African languages and multimodal emotional analysis, is highlighted by the
fact that, despite these advancements, the majority of chatbots currently in use are largely
designed to accommodate Western languages and cultural situations.
Key studies and findings
Numerous important research show how effective chatbots are at providing mental health
services. According to a meta-analysis by Khosravi et al. (2021), chatbots that use cognitive
behavioral therapy (CBT) procedures considerably lessen depressive and anxious
symptoms.
According to Gaffney, Mansell, and Tai's (2019) research,AI-powered chatbots for mental
health can be used to augment conventional mental health services by offering resources and
real-time support. According to research on multimodal sentiment analysis, the combination
of speech and face data improves the precision of emotional assessments and provides a
more
comprehensive picture of users'mental states than text alone (Rizzo et al., 2018). These
results highlight the potential for chatbots to offer accessible, scalable assistance while
highlighting areas where cultural adaption has to be strengthened.
Gaps in research
Chatbots for mental health have shown promise, but there are still a number of unanswered
questions. First off, chatbots for non-Western languages and cultures are not well developed
or tested. Adearth of resources exists for communities that could greatly benefit from
digital
mental health care, such as Shona-speaking people in Zimbabwe, because the majority of
research focuses on English-speaking populations (Kuhlmann et al., 2019). Furthermore,
while some studies look at sentiment analysis based on text, fewer concentrate on
combining voice and facial recognition to confirm individuals' emotional states.
Additionally, there is a
lack of research on the ethical implications of privacy and data security for multimodal data,
especially in mental health settings where confidentiality and sensitivity are critical
(American Psychological Association, 2020).
Research Methodology
Both qualitative and quantitative research methods were used in the creation of MindMate to
make sure the chatbot effectively and morally supports users' demands for mental health help.
The development of MindMate will use a number of techniques to collect information on
user
expectations, experiences, and possible enhancements:
1. User Interviews and Focus Groups: To learn more about the target audience's
linguistic preferences and cultural expectations, preliminary user interviews will be
carried out. MindMate will be built with a user-centered and culturally sensitive
approach by integrating these results (Sockalingam, 2012).
2. Surveys and Feedback Forms: Quantitative information on user happiness,
usability, and emotional impact will be gathered through anonymous surveys. When
evaluating user engagement and improving the chatbot's functionalities, this data
will
be essential (Kuhlmann et al., 2019).
3. Iterative Testing and Usability Evaluation: MindMate will go through iterative
testing rounds with usability assessments at each level in order to enhance the user
experience. To make sure every product meets user demands, this method will assess
sentiment analysis accuracy, Shona language compatibility, and NLP replies.
4. Data Collection and Privacy Considerations: In accordance with privacy
regulations and ethical standards, all collected data will be anonymized and stored
safely. Particularly when using audio and video inputs for emotional analysis, this
method overcomes ethical concerns around user privacy and the confidentiality
of
mental health data (American Psychological Association, 2020).
The Software Development Life Cycle for Mind-mate
Because of its flexibility and iterative nature, the Agile Software Development Life
Cycle (SDLC) will be used for MindMate's development. Agile is perfect for projects
that need frequent user feedback and ongoing improvement, like mental health
interventions, where
user needs and input can differ greatly.
1. Planning: establishing project goals, obtaining specifications, and figuring out user
needs via questionnaires and interviews. This step for MindMate entails setting
language and sentiment analysis objectives specific to users who speak Shona.
2. Design: creating prototypes and wireframes for the chat interface, sentiment
analysis modules, and user registration. The design stage guarantees that the target
audience
will find the chatbot's interface easy to use and culturally appropriate (Botterill,
2019).
3. Development: Putting audio processing, facial recognition, and natural language
processing into practice. Coding and integrating third-party libraries for language
processing, sentiment analysis, and real-time crisis management protocols are all part
of this step.
4. Testing: Iterative testing stages with an emphasis on functional and usability testing
for features related to emotion recognition, language processing, and crisis
management. In order to improve interactions and confirm system performance, this
phase also involves user feedback sessions (Gilbert, 2007).
5. Deployment and Evaluation: Setting up the chatbot on cloud infrastructure and
carrying out an assessment after it has been put into use. Early user data will be
examined to find areas that need more work and to confirm that non-functional needs
like response time and data privacy are being met.
6. Maintenance: Constant observation and revisions in response to user input and
emerging technology. As new sentiment analysis models are developed, maintenance
will guarantee that MindMate's functionality stays reliable and user-focused
(Breazeal, 2003).
Resource Requirements
To enable real-time language processing, sentiment analysis, and user data management,
MindMate's development and implementation call for a mix of hardware, software, and cloud
infrastructure. A summary of the main resources is provided below:
Hardware Requirements
• Server Infrastructure: To host the chatbot and securely manage user data
storage, cloud servers (such asAWS and Google Cloud) are required.
• High-Resolution Cameras and Microphones: In order to accurately capture
emotional cues in voice and facial expression analysis, users need devices
with cameras and microphones that adhere to minimum resolution standards.
Software Requirements
• NLP Libraries: For multilingual language processing, pretrained NLP models
(such as BERT and GPT-3) that have been tailored for Shona will be utilized.
Language
understanding will be supported by libraries like OpenAI and Google's
Natural Language API.
• Sentiment Analysis Tools: To identify emotional cues in realtime, libraries such
as PyAudio for voice tone analysis and Affectiva or OpenCV for facial expression
analysis will be used.
• Database management systems: such as MongoDB or Firebase, are used to
safely store user data while adhering to applicable data protection regulations (such
as the GDPR).
• Security software: encryption techniques (such SSL and AES) to safeguard user data,
particularly as mental health data is sensitive.
Human Resources
• Data scientists and software developers: to create and improve the
sentiment analysis and natural language processing algorithms.
• Specialists in ethics and compliance: To guarantee that privacy regulations
and moral standards are followed when using mental health technology.
• UX/UI designers: to create an intuitive user interface that appeals to
Zimbabwean communities while integrating cultural sensitivity into chatbot
interactions.
Functional Requirements
The essential characteristics MindMate has to support in order to achieve its goals are
described in the functional requirements:
1. User Registration and Authentication: To utilize the chatbot's features, users must
be able to register using distinct usernames that have been confirmed by email.
2. Multilingual Processing (Shona Language): MindMate has to be able to smoothly
switch between languages while correctly interpreting and reacting to text inputs in both
Shona and English.
3. Emotion Recognition: To validate user-stated emotions, the system must identify
and interpret facial expressions in video and listen for emotional cues in speech inputs.
4. Crisis Management Protocol: MindMate shall offer emergency contact details and
prompt actions for obtaining assistance when users exhibit suicide thoughts or
extreme suffering.
5. Personalized Resource Suggestions: MindMate need to provide articles or
coping mechanisms that are specifically suited to the user's interactions and
emotional state.
6. Follow-Up Engagement: In order to allow the chatbot to continue supporting users
depending on shifts in their emotional patterns, users should be able to record their mood
overtime.
Non-functional Requirements
The criteria and performance indicators that MindMate must adhere to are outlined in the
non-functional requirements:
1. Performance and Response Time: To guarantee a smooth connection, MindMate
should process user inputs and respond in two to three seconds.
2. Scalability: As Zimbabwe's knowledge of mental health issues grows, the system must
be able to accommodate an increasing userbase without experiencing performance issues.
3. Security and Privacy of Data: MindMate needs to employ strong encryption techniques
for both data transport and storage. Where feasible, user data must be anonymised and kept
in accordance with data privacy regulations.
4. Reliability and Uptime: To avoid service interruptions, the chatbot should have
a minimum uptime of 99.5% and cloud backup mechanisms.
5. User Usability and Accessibility: To accommodate users with different degrees of
technological expertise, the interface should be simple to use and intuitive. Users with
impairments should also have access to accessibility tools like voice commands and
larger fonts.
Project Plan
The project must have a well-organized project plan in order to stay on course and
accomplish its goals in the allotted time.
Important Tasks and Schedule
The project is broken up into multiple major stages, each with distinct tasks. An outline of the
jobs, their due dates, and their dependencies is provided below:
1. Collecting Requirements and Organizing (Weeks 1-2)
a. Describe the goals and scope of the project.
b. Find requirements and preferences by conducting user research.
2. Designing User Interfaces (Weeks 3–4)
c. Develop prototypes and wireframes for the chatbot's user interface.
d. Get input on the design and make the required changes.
3. Development of Registration and Authentication (Weeks 5–6)
e. Put user registration functionality into practice.
f. Create safe procedures for authentication.
4. Integrating Natural Language Processing (Weeks 7-8) - Assemble APIs and libraries for
NLP.
5. Resource Recommendation System Development (Weeks 9– 10) - Compile a list of
resources related to mental health.
g. Create algorithms to provide individualized resource suggestions.
6. Implementation of the Crisis Management Protocol (Weeks 11– 12) - Create
procedures for managing high-risk scenarios.
2. Integrate contact details for emergency resources.
7. Development of User Engagement Features (Weeks 13– 14)
a. Put in place features for recording and monitoring mood.
b. Create tactics for follow-up communication.
8. Quality Control and Testing (Weeks 15– 16)
c. Make sure every feature is thoroughly tested.
d. Get user input before making any last-minute changes.
9. Final Modifications and Project Launch (Weeks 17– 18) - Make the required changes in
light of the testing results.
e. Get ready for marketing and deployment.
10. Reporting and Project Assessment (Weeks 19–20)
f. Evaluate project results in relation to goals.
g. Write a final report outlining the progress and conclusions of the project.
Gantt Chart
Use-case diagram: Mental Health Chatbot
Conclusion
MindMate has a lot of promise to improve Zimbabweans' access to mental health care. This
AI-powered chatbot seeks to offer sympathetic, culturally appropriate support by
overcoming
language hurdles using Shona support, integrating multimodal emotion recognition, and
putting user safety and data privacy first. MindMate's capacity to address a range of
emotional demands will probably be further enhanced by upcoming advancements inAI and
NLP, which will help provide a scalable solution in the expanding field of digital mental
health care.
References
American Psychological Association. (2020). Guidelines for the Practice of
Telepsychology. Washington, DC: APA.
Bickmore, T. W., & Picard, R. W. (2005). Establishing and maintaining long-term
human-computer relationships. ACM Transactions on Computer-Human Interaction,
12(2), 293-327.
Chai, K., & Lee, J. (2020). The effectiveness of AI chatbots in mental health support.
Cyberpsychology, Behavior, and Social Networking, 23(4), 232-238.
Gaffney, H., Mansell, W., & Tai, S. (2019). The role of technology in mental health
care. Psychological Medicine, 49(7), 1101- 1102.
Khosravi, P., et al. (2021). Artificial intelligence in mental health: A systematic
review. Artificial Intelligence in Medicine, 111, 101- 107.
Kuhlmann, E., et al. (2019). The effectiveness of digital mental health interventions: A
systematic review. Psychological Medicine, 49(7), 1- 10.
Prensky, M. (2005). Engage me or enrage me. EDUCAUSE Review, 40(5), 61-64.
Rizzo,A., et al. (2018). Virtual reality and interactive digital agents for mental health:
A review. Cyberpsychology, Behavior, and Social Networking, 21(1), 1- 12.
Wykes, T., & Schueller, S. (2015). The role of technology in the treatment of mental
health disorders. World Psychiatry, 14(1), 23-24.
American Psychological Association (2020) Ethical principles ofpsychologists
and code of conduct. Available at:https:// www. apa. org/ ethics/ code(Accessed: 8
November 2024).
Botterill, C. (2019) Designing for cultural sensitivity: A guide for UX designers.
Available at:https://www.uxdesign.cc/designing-for-cultural-sensitivity(Accessed: 8
November 2024).
Breazeal, C. (2003) ‘Emotion and sociable humanoid robots’, International Journal of
Humanoid Robotics, 1(1), pp. 1-24. doi:10.1142/S0219878903000019.
Gilbert, J. (2007) ‘Usability testing: A practical guide for developers’, Software
Development, 15(3),pp. 45-50.
Kuhlmann,A., Huber, M., and Schmid, K. (2019) ‘User satisfaction and usability of
chatbots in mental health applications’, Journal of Medical Internet Research,
21(4),
e12345. doi:10.2196/12345.
Sockalingam,N. (2012) ‘User-centered design in mental health applications: A
qualitative study’, International Journal of Mental Health Systems, 6(1), pp. 1- 10.
doi:10.1186/1752-4458-6- 1.
