I am from a small city in Brazil, where the health facilities are not the best in the world. Twice in the past, my grandparents were misdiagnosed in the city's hospital, what caused serious health complications. Moreover, we all know that being sick can be an extremely vulnerable position both in terms of isolation and - in the hopes of obtaining an answer - susceptibility to misinformation. Considering this scenario, I wanted to build a solution to decrease false diagnosis and, moreover, provide the guidance patients need to take care of themselves, connect with people going through the same as them and, as a community, have a safe way to seek truthful information.
What it does
Health fox is an iOS app that provides support for both the doctor and the patient. Those are the functions available for the users 1. The Patient
- The patient has access to a daily checklist of thing he/she has to do in order to maximize the quality of the treatment at home. Items in the checklist might include the times to take specific medicines and vitamins, clean wounds, or go to doctor appointments. At each time the patient completes one item in the checklist, he/she will gain some reward points.
- The points can be exchanged to new animal icons, which the patient can collect and see information about them
- The patient has access to all their exams. When they click in one exam, they have access to an explanation of both what the exam does, what was the goal, and what were their results (once the doctor posts it).
- The patient can have access to a community of people that received the same diagnostics. He/she is able to post at this community and see other people's posts. Every post can also be seen by the doctor, who can analyze the general mood/mental health of the community and, if needed, take measures to prevent misinformation.
2. The Doctor
- The doctor will have access to the patients exams, rewards and posts.
- The doctor will have access to a page where, using Sentiment Analysis, he/she can see the analyze the patient's posts and see the general sad, anxious or conflict-related tone of the messages. That can be used for the doctor to be aware of mental health issues in the community and take the necessary measures.
- If this information is provided, the doctor has access to a screen that analyzes the genes of the patient. It takes all genes with polymorphisms and looks for interactions between them. The genes are given scores based on their interaction in the network and the most represented biological pathways are shown at the top of the screen.
How I built it
I build the app in Xcode with objective-c language. For the sentiment analysis, I used databases of words with different contends (sad, anxious, etc) and crossed that data with all the posts in the network For the gene analysis, I used data from GeneMANIA to find the interaction between the genes and the most represented biological pathways. Moreover, I computed the degree (number of connections) and betweenness (see below) of each gene in order to find the most relevant ones.
Challenges I ran into
I definitely had a challenge with time. I had to decrease the amount of things I planned to do for the app and go with a more simple design. However, I believe I was able to accomplish the most important topics I had planned to. Moreover, I had some problems with protocols in Xcode that were not working at first. After some time, however, I found the bug and got them to work. The last challenge was to find a clean and good way to show the gene interaction network. At the end, for the sake of simplicity, I chose a table and focused on the most relevant genes.
Accomplishments that I'm proud of
- Finishing everything on time
- Creating the gene network
- Being able to create some sentiment analysis
- I am more happy than I expected with the design of the app
What I learned
I believe I learned a lot about both genetics and app development. Moreover, I learned to better prioritize tasks for the sake of time.
What's next for HealthFox
I wish to improve the sentiment analysis and try another method of visualization of the genetic network. Moreover, I wish to implement more parameters that could help the diagnostics beyond the genetic network, such as a computer vision model to find brain abnormalities in MRI's or a simulation of the body to predict the most probable place of nerve damage given the symptoms.