Opportunities with AI


Any article on AI that is written and published in 2023 must begin with a clarification: none of this was written by ChatGPT. Not to say that there would be anything wrong with sections being researched or edited with AI; simply that we should be transparent about this.

I am very excited about the opportunities being opened by AI. Last August I spent time with DALL*E, an AI artist. Below the outcome of a prompt that I gave, “Oil painting by Matisse of a child wearing a virtual reality headset in an examination room in a hospital.” Looking back, ChatGPT could have probably given a more refined textual input to DALL*E.  

Medical Doctors all over the world are faced with many of the same basic challenges that AI will hopefully help us with in the future. At the heart of my work, and the work of my colleagues, is the patient. And with the patient we form a partnership, often with the goal of managing a disease or returning them to good health.  

A big part of my job is therefore to help a patient understand the diagnostic workup that we do, and then the results and corresponding treatment options. Here AI has great potential to generate much better documentation that the patient takes away with them, specific to their situation and treatment plan. I certainly would see it as an aid that helps this partnership with the patient.

Yet still, if I look at a typical day, some of my time is spent on activities like finding the address of the referring doctor that I need to write a letter to. These tasks shine a light on the opportunity to systematically collect, store, and retrieve data across the healthcare network. In our current state of disconnected information, the patient is often the gatekeeper of their own health history, and perhaps digital solutions can first and foremost help the patient to take ownership of the many different interactions they have with healthcare professionals.  

Of course, as a doctor, I want to have access to the relevant parts of this healthcare data to make the best diagnosis. AI has the potential to help this partnership of the patient and the doctor by highlighting data that is abnormal, not just for a typical patient population, but for that specific patient. Part of our accumulated experience as a doctor is knowing which data are relevant for a diagnosis, and summarized data that highlights abnormal parameters would help to democratize this specialist knowledge.  

One final thought is to do with the for standardization of clinical routines. Most healthcare systems that I’ve worked in strive for greater efficiency, but we must be conscious that the most efficient clinical routine changes based on the rapid advancement of scientific research leading to new diagnosis and treatment options. There is no “perfect” solution for healthcare, simply “optimal” solutions given the state of knowledge and technology. Part of our use of AI will be to assimilate new information and support us in our iterative optimization of the clinical routine. AI can help us with administrative and routine tasks, allowing doctors to spend more time with their patients. Which, in my opinion, should be the goal of every physician.

I am very grateful that at machineMD we have a team and partners that are embracing the opportunities of AI with our stated mission to radically improve the early diagnosis of neurological disorders.