HealthMatch

Helping People get the Help They Need.

Comment

Inspiration

When our team thinks of creating something that impacts humanity, the first thing that comes to mind is, how can we make the lives of the people we love better?.

Our project was inspired by our teammate's health journey during this past year. She experienced a major health set-back that left her unable to walk for six months. Three of those months were spent running from doctor to doctor, making little progress, trying to understand what was happening to her, and how she could fix it. If she had known which doctor, or even which section of medicine, would be able to help her, we are confident that it would not have taken nearly as long for her to walk again.

While her story is an extreme case, the Global Burden of Disease study reports that over 95% of the world's population has health problems (1). They additionally report that 2.3 billion individuals experience more than five ailments at any one time (1). To make matters worse, Johns Hopkins Medicine estimates that significant permanent injury or death annually in the United States ranges from 80,000 to 160,000 American's suffer permanent injury or death annually due to missed diagnoses, delayed diagnoses or flat out incorrect diagnoses (2).

HealthMatch was created with the intention of minimizing missed and delayed diagnoses, striving to help people streamline their healing journey.

What it does

HealthMatch Helps Patients Get the Help They Need.

HealthMatch is a one-stop shop for determining which specialists to consult for your symptoms, finding a reputed specialist near you, and obtaining information to schedule an appointment. It works by prompting the user to enter their symptoms, which they can do in whatever format they prefer, whether that be bullet points, a list, or even a normal sentence. From there, the user is prompted to enter their zip code, and can opt to narrow their search down by healthcare system.

Once done entering their queries, the user is presented with the relevant medical specialties, along with the definitions of these specialties. The user can select one of these categories, which will then take them to a page of nearby doctors who specialize in that field. Here, the user can see the doctor's name, photo, and city of practice, and if they select a doctor, they are taken to the doctor's webpage on their practice's website. The user should be able to schedule an appointment with their chosen doctor, through that provider's own system, in order to not cause conflicts or introduce any chance of contact with privileged data.

How we built it

Our front-end is powered by HTML, CSS, Javascript, and Figma and our back-end features Python, Flask, as well as the use of the DaVinci 3 AI Model and the Yelp API.

HTML HTML was used for all text manipulation that occurs on our website. This includes our suggestions for queries, headings and product summaries that appear on the site.

CSS CSS was used to make the elements of our website match those that were laid out in our design.

JavaScript JavaScript was used to write the entirety of the behavior executed by our webpage. In addition to this, we used JavaScript to send and receive data to/from the back-end of our project.

Figma Figma was used for the design process of our website, slides, and logo. Serving both as a useful communication tool, and as a creative tool, Figma powered all of the designs that you see in our project.

Python Python was used to use the API calls necessary to access the Davinci 3 AI Model and Yelp data. Once called, Python was used to parse through the data that we received, in order to make it match the format that front-end expected to receive the information in.

Flask Flask was used as a framework for Python to receive data from all of the different APIs being called, and to communicate that data with the front-end.

Davinci 3 AI Model The user-entered symptoms are sent as queries into the DaVinci 3 AI model* through the public APIs, where we are able to obtain the specific branch of medicine the symptoms fall under. The medicinal specialties are then parsed, defined, and sent to the front-end, using Flask as the framework to do so.

* We carefully trained our AI to never provide medical diagnoses, suggestions of surgery, or medical advice. We kept a very tight leash on the information exchanged between the two, in order to preserve the safety and privacy of the user.

Yelp Fusion API Once the user chooses the specialty they want to go ahead with, that choice is then queried back to the backend, along the zip code and healthcare provider they chose, which are used to parse through the Yelp API to find doctors and clinics in that are accessible to the user.

Challenges we ran into

Our team is not very experienced in React! While we tried for a very long time to utilize this framework, at the end of the day, we didn't have enough experience to properly execute the front-end to back-end connection using this framework. However, our team had enough experience in pure JavaScript that we were able to pivot and pull off the connection regardless.

Accomplishments that we're proud of

This hackathon had a lot of subversions of expectations for the entire team. We ended up doing jobs that didn't overlap at all with what we assumed we'd do, working with people we've never worked with before, and creating the most fully fleshed-out product that any of us have ever made for a hackathon! From the front-end, to the back-end, looking at the functionality, scalability, and visual designs, to the future scope and business plan, we are proud of everything we've put into this project!

What we learned

This was our team's first time working with AI Models, which was a much-anticipated experience, but a learning curve to be sure. Now, after spending so much dedicated time working with one, we now feel much more confident using them, and look forward to implementing them in our personal projects in the future.

Additionally, this was some of our designers' first time designing the branding for a project, which was a fun task to try to fulfill, and we feel that we could readily apply the lessons learned here in the future.

On top of that, members of our team with less coding experience were able to learn so much and contribute so much to this project, that it's hard to summarize it all. Starting out with minimal experience in React, and then spending the majority of the Hackathon working in pure JavaScript on the integration of the front-end and back-end was a difficult, but rewarding experience.

What's next for HealthMatch

While our product certainly stands on its own, we recognize that it primarily serves those patients who find their doctors via a search on the web, not those who look on their usual providers website, and isn’t very accessible for older people.

We see a future where HealthMatch exists both as a standalone web-application and as an integration with large healthcare systems, such as Sutter Health, Stanford Health, or UCSF. We made sure to prioritize creating a mock-up of this vision, which can be found in our photos. We believe that this vision would allow us to reach those patients who we cannot reach on our own.

Inspired by the Epic Software business model, which currently powers the user portals for the majority of large healthcare systems, we see this happening via selling an integration of our product's capabilities to these large healthcare systems. We believe that we provide great value to healthcare systems by saving them money by freeing up doctors' time, automating patient assistance to a degree, and assisting in patient retention. Our implementation of our functionalities would change with this business plan, no longer relying on the Yelp Fusion API, but drawing doctor data from the systems' own internal database of providers.

If people know who to turn to for help, then their health journey will be streamlined. With less guesswork, doctors won’t get as many patients that they are unable to help. HealthMatch allows doctors to help more people, and for more people to get the help that they need. In short, we believe that teaming up with these healthcare giants will allow us the reach to make a tangible impact by saving doctors’ time and allowing them to see more patients who really need their help, all while helping patients get help faster.

Citations

(1) The Lancet. "Over 95% of the world’s population has health problems, with over a third having more than five ailments." ScienceDaily. www.sciencedaily.com/releases/2015/06/150608081753.htm (accessed February 19, 2023). (2) Toker, David Newman, Ali Shabahang Saber Terani, HeeWon Lee, Simon C Mathews, Andrew Shore, Martin A Makary, and Peter J Pronovost. “Diagnostic Errors More Common, Costly and Harmful than Treatment Mistakes - 04/23/2013.” Johns Hopkins Medicine, based in Baltimore, Maryland, April 23, 2013. https://www.hopkinsmedicine.org/news/media/releases/diagnostic_errors_more_common_costly_and_harmful_than_treatment_mistakes.

Built With

Share this project:

Updates