mPath

The Python prototype used natural-language processing to identify useful details in a free-form symptom description and convert them into structured input for the triage flow. We chose this interaction because people rarely describe how they feel in the same vocabulary as a medical form, especially when they are already unsure which kind of care they need.

The parsing stage had to preserve information related to urgency without inventing detail from vague wording. For the hackathon demo, we kept the extracted information visible in the flow so the user could see how their description affected the recommendation.

We combined the NLP pipeline with web scraping and the Google Sheets API to support the assessment and map its result to an appropriate level of care. The team kept the mapping in a shared Google Sheet so we could revise it quickly as the product flow changed during the hackathon.

The recommendation pointed toward a level of care a person should consider and made no diagnostic claim. A deployable healthcare product would require clinical review, validated decision rules, privacy safeguards, and a much more extensive evaluation than a hackathon prototype could provide.

mPath placed in the top 10 out of 30 teams. We documented the user problem, technical flow, and proposed path forward in a pitch deck and one-page summary, then used the working prototype to demonstrate how a symptom description moved through the system.

For the final demo, we entered a symptom description, showed the structured information produced by the NLP step, and followed it through to the care recommendation. The presentation also listed the clinical review, validation, privacy work, and ownership model that would be required before a system like this could be used outside a hackathon.