This article picks up where the previous one left off, How We Built an AI Startup in a Weekend Hackathon in Germany, focusing more on the final product rather than the hackathon process itself.

When you're creating an AI product about health, there is always a question of ethics, accuracy, responsibility, and trust. That's why it's important to separate healthy lifestyle products and real medical ones. We thought long and hard about the line between them and how close we can get to it. It's a complex question, but we found a solution: create two standalone products that are far away from the line but on opposite sides.

"Best Friend"

The first one is about a healthy lifestyle. It's your virtual best friend, and even more: assistant, coach, motivator, sometimes even your mom or a mini psychologist. By collecting essential information about you from the chat and by having in memory your anamnesis, bio, and history, the best friend can just listen to how you spend your day, comment on it, help you with everyday advice and suggestions, or even help you move towards your goals and dreams using personalized motivation, human sympathy, and understanding your feelings.

"Dr. House"

Problem

The second product is a real medical one. Let's begin with the problem:

• The complexity of the diagnostic process itself due to the complex structure of the human body and the huge number of possible cases that cannot always be completely treated by a doctor
• Overburdened healthcare systems and professionals
• Inefficient and delayed diagnosis in healthcare, leading to worsened patient outcomes.
• Accessibility issues for remote or underserved populations
• People's reluctance to visit hospitals, fear of doctors.

Diagnosis is important

Why is diagnosis so critical? Einstein is quoted as having said, “If I had an hour to solve a problem, I'd spend 55 minutes thinking about the problem and five minutes thinking about solutions.” The point he makes is significant: preparation has great value to problem-solving. The same holds true in medicine. Professionals say: correct diagnosis is ~70% of healing. And we believe we can improve 70% of modern medicine via a single AI product.

Implementation

Meet Dr. House — AI diagnostician in your phone. A mobile app that provides AI-powered diagnosis in minutes.

Now, how are we going to achieve this? From the beginning of time until now, we believe that the best way of communication is speech. If you want to know something — ask questions. And by asking questions, I mean asking correct questions. So, we're going to make an app that will have your full anamnesis and will ask correct personalized questions and based on the answers suppose possible diagnoses. Sounds pretty simple but helpful, isn't it? And much better than googling some of your symptoms (I have cancer every time).

And it helps not only ordinary people. Doctors can also use this for double-checks, second opinions, or preliminary prognoses, for example in an ambulance.

Also, when Dr. House recommends a visit to the hospital, the process will be significantly faster because the app already has complete anamnesis, medical history, and a filled questionnaire, so both the customer and doctor need to spend significantly less time. And the same app can be used as an emergency card if the user is “offline”.

Can do better?

Can it be even better? We say yes! How? Integration! The app can also take into account info from Apple's HealthKit and medical devices. Or vice versa, the API can transfer data to the hospital (and vice versa) even without direct human contact.

If the app monitors one's health, it means the person can be sent home from the hospital much earlier and free up space for somebody who requires it more. Doctors will monitor the status remotely, and the app will notify them if something is wrong.

Also, if we want ordinary people to use this app, we should add a better description of diseases using simple words; translate from medical to English.

Conclusion

Unfortunately, our team does not have the opportunity to work on this startup full-time, so we published all the results of our work in the public domain. You can find the slides for this article here . For full details on the project, including the source code and the custdev, check the GitHub repository.