The dawn of AI for common use has begun with GPT after many stumbles from simple chat bots to recommending cancer treatments. What the printing press did to preserve knowledge, the internet did to connect humanity, blockchain showed how to program trust, the current GPT models are doing to accelerate learning. Exciting times indeed, but to what end, what real-world problems can AI solve for us?
Looking at some of the most challenging problems of our times there seems to be an opportunity for making the most impact in Healthcare starting with solving problems of inequity. With a growing demand and constrained supply of facilities and personnel for Healthcare, the current model is unsustainable due to high costs or lack of access. For AI to be able to make a dent in the quality of care by making it independent of location would be a huge first step.
With AI, the limitations of a one size fits all approach can be overcome with personalized research before making a diagnosis or recommending treatment plans. Using trained learning models to level the playing field that allow a doctor in a small rural location to have the same access, to diagnosing an uncommon condition, as that of an urban specialist would help lower costs. The advantages would run both ways, imagine a specialist in a research and teaching hospital learning from traditional medicines and regional therapies. Having a true digital assistant with an easy to use chat or voice based command interface means no longer subscribing to expensive knowledge base that take a PhD to decipher or waiting for consulting specialists time. Sure, this has liability and deregulation implications, but it’s a start.
Solving access issues that started with tele-health, accelerating during the pandemic, are now poised at an inflection point. Imagine consulting with the best doctors in the world who already have your summarized information. A patient could share their symptoms, the GPT model would summarize, ask more questions and suggest relevant inputs with patient monitoring data. Now during the tele-health visit the homework is done and the 15 minute consult can be much more productive. Real-time video based biometrics can confirm the patient history, assess current condition, live translation can remove barriers and a post-visit recorded summary can help guide future treatment.
AI can help drive efficiency and access with digital assistants, that democratize healthcare delivery. Similar approaches to prevention with personalized regimens and machine learning based diagnostic recommendations in medical imaging would help the continuum of care. There maybe enough resources being spent on healthcare but deploying them in a more equitable manner is where AI can have the most impact.
Rad@Sami 