[https://github.com/HitanshBhatt/UofTHacksX]
Inspiration
Any academic setting naturally comes with stress and anxiety. We have personally witnessed some of our close friends struggle to speak out for help and guidance when evidently struggling with mental health issues, only because they weren't sure if there are any resources available, and if there are, how to access them. Some are more comfortable than others when speaking up for themselves and seeking help. Provided below are consequences of inadequate access to mental health services:
- Access: About 28,000 youths in Ontario were on waiting lists for mental health treatment in 2020; doubled since 2017 [1]
- Stigma: In a 2019 survey, 75% of respondents expressed reluctance to disclose a mental illness to coworkers [2]
- Morbidity: The disease burden of mental illness in Ontario is 1.5 times higher than all cancers put together and more than 7 times that of all infectious diseases [3]
- Mortality: Mental illness can slash 10 to 20 years from an individual's life expectancy [4]
- Economic Loss: $50 billion per year is the estimated annual economic cost of mental illness in Canada, related to loss of productivity and reductions in quality of life. This accounts to around ~2.5% of Canada's GDP [5]
We see this as a major problem and decided that we want to create a platform for such people where they can get direct access to existing mental help resources.
What it does
MentalBase is a platform where you can get all your mental health resources at one place. You can anonymously talk to an Artificial Intelligence bot, or if you're comfortable, an agent who will not know who you are unless you choose to disclose that information. You can type in any concerns to the bot and it will provide you with a list of resources that match your search query. Furthermore, if you're seeking for more human-like responses, or are just looking for someone to talk to, you can chat with an agent and they will help you to the best of their abilities. The bot and the agent will provide resources for mental health that are offered particularly to university students to better cater your needs. If you feel comfortable sharing your information, the bot will run through your responses to provide you extra information like resources closer to your physical location, tailored mental health services, etc.. We decided to solve the problem of allowing such students to open up without having to disclose any of their personal information unless they voluntarily provide it to better their experience.
How we built it
The major component to getting information about existing mental health resources was to scrape websites that provided such information. We used Python to build an algorithm to scrape data from websites and store it in our database. The data is fed into our NLP (Natural Language Processing) algorithm to train the chatbot AI to generate responses to specific queries. After training the AI and testing, we linked the frontend (our website) to the backend (Python scripts) using a server hosted on Flask.
Challenges we ran into
Scraping data from different websites was a challenge since each website had a different structure. Hence, in order to extract data, a new algorithm had to be programmed for each website.
Accomplishments that we're proud of
The ability to build a ChatBot, learn and use NLP to develop an efficient search algorithm such that user mental health requests can be catered. Furthermore, we scraped data from websites despite varying HTML structures. As such, concise summaries of mental health services were concluded.
What we learned
We learned how to use web scraping tools (Beautiful Soup), acknowledged its capabilities for efficiency by appreciating the ease of acquiring important information from the Internet. Adding on, we gained the knowledge to use NLP to conduct semantic searching. Likewise, we familiarized ourselves with the web framework, Flask and its advantages in creating flexible code, API support, etc. to interface both the frontend and backend.
What's next for MentalBase
We want to expand our platform by automating certain aspects such as web-scraping to make the backend side of our platform more efficient and accurate. We also want to tune our NLP model further to suit our needs better as we were only able to provide a small set of training data in the given time.
References
[1] Statistics Canada, 2020
[2] Ipsos (2019). Mental illnesses increasingly recognized as disability, but stigma persists. Retrieved from https://www.ipsos.com/en-ca/news-polls/mental-illness-increasingly-recognized-as-disability
[3] Patten et al. (2005). Long-term medical conditions and major depression: strength of association for specific conditions in the general population. Canadian Journal of Psychiatry, 50: 195-202.
[4] Chesney, Goodwin & Fazel (2014). Risks of all-cause and suicide mortality in mental disorders: a meta-review. World Psychiatry, 13: 153-160.
[5] Smetanin et al. (2011). The life and economic impact of major mental illnesses in Canada: 2011-2041. Prepared for the Mental Health Commission of Canada. Toronto: RiskAnalytica.
Log in or sign up for Devpost to join the conversation.