Artificial Intelligence (AI) will not replace physicians, but physicians who use AI will replace those who do not.
– Eric Topol, physician and influencer
Recently, a patient came to see me for an eye abnormality relating to Diabetes. His family doctor had utilized an AI-enabled handheld device, which in turn connected to a smartphone, and had performed a retina examination. This device found some evidence of retinal disease in his eyes. This assessment was important because the patient had Diabetes. The retina is a screen that sits at the back of our eyes with which we can see the world, and diabetic retinopathy is an important cause of future blindness. Our team did a quick evaluation. And yes, his eyes had sight-threatening retina disease, which we promptly managed.
This is illustrative of many such cases where AI has helped detect and manage diseases. As I argued in a medical journal some time ago, artificial intelligence-based assessment of the retina is the clearest elucidation for understanding AI in healthcare. “Human” intelligence observes, analyses, makes conclusions and takes action. A smartphone based retinal scan observes the retina, analyses the lesions, draws a conclusion about whether the retina is okay or not – and finally takes action, i.e suggests the need for advanced evaluation, thus displaying “artificial” intelligence.
Not just doctors, patients are also now using AI for Diabetes care. A real example I came across is a young child with type 1 Diabetes. Often these children have to literally count and calculate their carbohydrate intake for dosing insulin. This patient and her parents were using an AI application to quickly calculate the carbohydrate content and decide on insulin doses. Just take a photo and calculate the carbs! Indeed, patients are also teachers for us, and after that incident I regularly encourage patients to use AI for this purpose, but to take insulin only as per our Diabetes team’s guidance. The guidance is very important because, sometimes, AI tools can hallucinate and an inadvertent dosage of insulin can lead to catastrophically low glucose levels.
From radiology to robotic surgery, and from heart health to metabolic care, AI use has been welcomed by doctors and patients alike. Wearables today can detect cardiac abnormalities. Recent AI-enabled insulin pumps, which can measure and even predict glucose levels and help to calculate insulin dose appropriately, are excellent at keeping the blood glucose in the Goldilocks Zone : not too high, not too low, just within the right range.
This issue’s cover story will guide you through the dazzle of AI as well as its under-appreciated dangers. A celebrated doctor and an astute patient with Diabetes contribute their perspectives to this issue of DH. AI is revolutionary but like any technology, it has its downsides too.
“A good human plus a machine is the best combination,” said Garry Kasparov, the former world chess champion. This more or less sums up the current situation, for most AI experts agree that the human-in-the-loop is the most important aspect of using AI.
Happy reading! And as you read, remember that even though healthcare in the 21st century may be all about technology and AI, the real intelligence that drives it, the hands that heal, the heart that cares, will all remain human!
