The current COVID-19 outbreak made us realize just how much pressure our healthcare providers are under. One of the participants in our group knew of a CFS specialist who was interested in a product that would save his time in diagnosing CFS. After doing some research and validating the usefulness of our product in a CFS clinic, we decided to focus our project on how our product could be used to save doctors’ time in diagnosing CFS.
What it does
Our chatbot is introduced after a patient has been referred to or after they have booked an appointment with a Chronic Fatigue Syndrome specialist. The chatbot asks the patient what symptoms of CFS they have, based on the CDC guidelines for diagnosing CFS. The patient’s symptoms and medical history get integrated within the Electronic Health Record system of the hospital and allow the physician to see the patient’s information prior to the appointment with the patient. This will save doctors’ time during the consultation and reduce their stress.
How we built it
Challenges we ran into
While dealing with the data retrieval, we had to manage the ethical implications of our chatbot. The main challenge was about how to keep the patient’s information confidential and secure. We talked to a Senior IT Architect at an NHS Trust, who suggested that we use Health level 7, a standard method of data transfer used by healthcare systems, to integrate DekomAI with the EHR. By integrating our chatbot with the hospitals’ medical systems using HL7, patient privacy is automatically ensured via the hospital’s security system. We also struggled with deciding on the exact function of the chatbot. We originally thought of having the chatbot perform differential diagnosis, where it ruled out other conditions similar to CFS and then told the patient whether they should or shouldn’t visit the doctor. However, we decided that even if the patient didn’t have CFS according to the diagnosing criteria, the patient should still see the doctor. Also, after conducting our user surveys and speaking to 12 doctors, most of them didn’t want the chatbot to diagnose CFS itself. So, we decided to create a chatbot to record symptoms and act as an aid to physicians instead.
Accomplishments that we're proud of
Throughout this hackathon, we have built Dekom.AI from the ground up. We are proud of the potential that our chatbot has in helping our healthcare providers, specifically during this period of crisis. We are also proud of all of the collaboration that went on in our team, despite having people from three different time zones. We are proud of how quickly and efficiently we were able to divide up tasks in order for us to accomplish what we did. Lastly, we are proud of and grateful for having the chance to spend our time doing something meaningful during this pandemic, which could potentially help our workers on the front lines.
What we learned
During this week we have learned invaluable skills. FIrstly, we learnt to collaborate with strangers and use our various strengths to help the project go forward. Each and every one of us will come out of the hackathon having learnt equally about leadership and teamwork. Also, while making our pitch deck, we learned how to narrow down our ideas and efficiently summarize our project. Furthermore, we learned how to build off of others’ constructive criticism in a short amount of time. We also expanded our knowledge about the way healthcare systems work in different countries and how our product might integrate into these systems.
What's next for Dekom.AI
Our prototype of Dekom.AI receives ‘yes’ or ‘no’ answers from the user about a set list of symptoms that are associated with Chronic Fatigue Syndrome, and if the user responds ‘yes’, Dekom.AI asks further questions about the severity of those symptoms. In the future, we would like to integrate Dekom.AI with electronic health record systems so that it has access to information like medical history and family medical conditions, which would be tedious for a user to give to the chatbot at one time. This would better economize the user’s time. We would also work to get ethical approval from the NHS and the FDA so that we are able to pilot and test Dekom.AI in hospitals. Additionally, we want to use machine learning to teach Dekom.AI how to diagnose CFS, with options for the clinicians to give feedback on Dekom.AI’s accuracy, which would help us further improve our product. At that point, we would also like to expand Dekom.AI’s capabilities to other diseases, so that it’s able to assess a broad range of diseases rather than just CFS.