Prior studies have shown that bedridden patients are suffering from significantly high risk of mental issues due to the isolation from the community, while few effort has been spent on resolving the need of social connection of bedridden patients in the hospitals. In addition, the support system formed by family and friends of patients may not be available 24/7. In this hackathon, we dicided to tackle this problem by creating Patient Talk, a semi-annoymous communication channel for patients sharing similar conditions, which alarms doctors of patients' early signs of mental problem while maximizing the privacy of patients.

What it does

Patient Talk is formed by communication groups around medical conditions. Patients can join the groups of their conditions and the groups serve as a supplemental support system, where patients can talk to other similar patients anonymously. In addition, patients can also log their thoughts in a private channel which serves as a diary.

Moreover, Patient Talk uses cognitive analysis to understand the mental condition expressed in the dialogue. As dialogue serves as a safe space for patients to communicate, any offensive messages are directly filtered. Moreover, patients can turn on the "safe mode" feature in the app which temporary hide overly-depressing messages in the groups. When Patient Talk detects that a patient has been expressing early signs of mental conditions, it will notify the doctor through email, without leaking the dialogue of patients to doctors.

Patient Talk also provides doctors with a dashboard to see the overall mental condition of their patients in the hospital.

How we built it

We use IBM Watson's emotion and sentiment analysis APIs to capture negative (especially abnormal level of sadness, anger, and fear) emotions in the dialogue. We use Microsoft Azure's content moderation APIs to filter sexually explicit or offensive messages to create a safe environment. We use firebase as our data storage solution. We use Flask / react to set up the core chatting service.

Challenges we ran into

Balancing patient privacy and information provided to medical expertise.

Accomplishments that we're proud of

Finished a minimum viable product with several different services integrated in a limited timeframe.

What we learned

Development planning and skillset balancing in an agile team is very important.

Share this project: