A couple of days ago, a user posted on X that Grok correctly diagnosed her daughter’s wrist while the initial opinion of doctor was incorrect. The user explained that one of her daughter was in a bad car accident where she escaped relatively unscathed. However, the pain in her wrist suggested something was amiss – maybe a fracture. The doctors and radiologists who examined the X-rays concluded that there weren’t any breaks and sent her home with a painkiller. The patient’s condition however got worse, she was unable to move her thumb and felt tingly. The worried mom started researching online. Remembering a post by Elon Musk that Grok 2 could read medical images, she uploaded daughter’s x-ray and asked if there were any abnormalities. Grok 2 replied “There’s a clear fracture line in the distal radius.” The user double-checked again giving the benefit of the doubt to doctors who said that the line in X-ray was growth plate. She asked Grok whether it was a fracture line or fused growth plate to which the reply was “It was fracture line”.
After double-checking with Grok, the patient’s mother next went to the doctor and got a referral to Ortho. A wrist specialist examined her this time and confirmed that Grok’s diagnosis of distal radial fracture was indeed correct. She was also told that her daughter would have needed a surgery if that displacement was left untreated for long. Even though the X user didn’t share the Grok diagnosis with the wrist specialist to obtain an independent assessment, it’s the Grok 2 results that prompted immediate action and avoidance of surgery. The doctors who treated her at first surely got it wrong while an AI tool alerted the patient’s mother in time to avoid medical complications.
When the X-user recounted her experience, the tech billionaire Elon Musk saw it as another PR opportunity for Grok and reposted the tweet. The age of AI clinical assistants is arriving and incidents like these are a prelude to the future.
Will AI’s diagnosis replace judgment of Clinical Assistants?
Clinical assistants are an integral component of health care systems worldwide. Assisting a doctor in real-life cases is a significant step in mastering medical symptoms. Clinical assistants are important in initial diagnosis because they often have as much experience as doctors in terms of number of medical cases. A clinical assistant recording vital signs of hundreds of patients per day acquires a keen sense of the capabilities of medical equipments, the symptoms of patients, and possible treatments that doctors may prescribe. Clinical assistants in-charge of medical tests and routine tasks gain expertise comparable to physician assistants. They provide care to the patients and interact with them more than the doctors themselves. Apart from assisting, the clinical assistants are the first line of contact after conducting a medical test or procedure. The bias or expertise of clinical assistants can unconsciously affect the judgment of the physician. The medical assistant’s play a key role because patients often turn to them for initial opinion right after a test whether it’s an MRI scan, X-ray, sonography or just a blood test.
In the given instance reposted by Elon Musk, the clinical assistant conducting the x-ray and doctors treating the wrist injury clearly got it wrong. They misunderstood the serious fracture line for a growth plate and sent the patient home with just an ibuprofen. The alert mother double-checked it when the condition and pain got worse. She was wise enough to seek assistance of AI but also got a referral to an orthopedic doctor instead of sticking to the first opinion of doctor. If not for a consistent and confident response from AI, she may not have taken the step of securing second opinion. A later discovery would have necessitated a terrible wrist surgery. AI saved the day when clinical assistants and doctors didn’t recognize the fracture.
In the future, clinical assistants themselves may complement their judgment with an AI tool. Medical devices may have built-in AI tools so that probability of wrong medical findings are reduced. Clinical assistants serve a larger purpose than merely assisting doctors. They provide service to the patient. They comfort them and look after them and carry out tasks determined by the physician. Most of their activities involve human interactions for which an AI chatbot tool may be inadequate for. An AI robot that replicates the human touch provided by clinical assistants will be the future especially in countries where there is an ageing population and human clinical assistants are in short supply.
Conclusion
When Elon Musk reposted the tweet, radiologists and doctors themselves were quick to point out their experience with AI tools like Grok. Several medical practitioners revealed that their experience with AI tools suggested that it got it wrong more often than right. The X post also saw counterclaims with anecdotes where AI chatbot diagnosed medical images incorrectly. It’s possible that AI tools haven’t yet attained that level of precision where their medical opinion can match that of expert doctors. It’s conceivable that specialized AI tools solely focused on medical images and reports alone can replicate the medical diagnosis of trained doctors. It doesn’t take away from the fact that AI tools are fast gaining expertise because it taps into vast number of medical reports. A medical practitioner who treats a hundred patients a day is limited by his exposure to various ailments, limited by his memory and processing capabilities. A medical chatbot or AI tool has the potential to process inputs of millions of medical reports and develop capabilities and experience vastly superior to human doctors. AI is catching up with clinical assistants and doctors even though they can never completely replace the component of the human touch.
