SMAS | Devpost

Inspiration

Nowadays, people put great emphasis on spending time as efficient as they can. Same applies to services, such as healthcare, where time means money and of course, healthcare quality. We came with an idea of appointing patients to an appointments with added values of machine learning, which we use to predict confidence level of coming to a scheduled appointment. This can make the healthcare more efficient by reducing gaps in schedules.

What it does

We built a WPF application where nurses can make schedules for patients effectively by predicting confidence level of attending. By having a number of patients ordered for each day and seeing if they come or not, healthcare providers can use SMS-service to inform next-in-line patients and fill the gaps and even possibly identify spikes and gaps for each day of the year.

Challenges we ran into

We had to implement No-SQL MongoDB database, since we did not manage to sucessfully create a cloud SQL database at Azure nor Google Cloud. This changed our plans, since we wanted to use SQL Azure database, which has really good support with C# and machine learning in Microsoft AzureML studio. For prediction, we have used an open-source dataset (https://www.kaggle.com/joniarroba/noshowappointments/version/5), which contains inconsistent data, so the prediction model is not very accurate.

Accomplishments that we're proud of

We mananged to work as a team and created a prototype of a product, on which we can present our idea. And of course, we are happy that we had fun at our first ever hackathon and survived this event.