Consumers sceptical of AI blood tests

A blood test is supposed to be a snapshot, not a fortune teller, but a growing list of consumer apps now promises to turn a page of lab numbers into a personalized health plan with artificial intelligence. Mashable’s April 10 review says those services can cost a few hundred dollars a year and still may not give users the clear answers they expect. (mashable.com) The pitch is easy to understand because most lab reports arrive before a doctor calls, and the reports are packed with ranges, flags, and abbreviations that look like a spreadsheet from another planet. American Medical Association chief executive John Whyte told Mashable that doctors are not always great communicators, which helps explain why people will pay for a translator. (mashable.com) What these tools usually do is simpler than the marketing sounds. They take numbers like cholesterol, glucose, or iron, compare them with reference ranges, and turn that into plain-language notes about diet, exercise, sleep, or whether to ask a clinician a follow-up question. (mashable.com) That can be useful in the same way a map legend is useful. It helps you read the symbols, but it does not tell you why traffic is bad on your street today, because blood work depends on timing, symptoms, medications, age, sex, medical history, and sometimes whether you even fasted before the draw. (mashable.com) The scientific problem is that a “normal” range is not a diagnosis. A value can sit just outside the lab range and mean very little for one person, while a result inside the range can still matter when it changes sharply from that person’s usual baseline. (myadlm.org) That is why laboratory groups are not rejecting direct-to-consumer testing outright, but they keep adding guardrails. The Association for Diagnostics and Laboratory Medicine said in its 2024 position statement that only reputable Clinical Laboratory Improvement Amendments-certified labs should perform this testing and that consumers should have enough expert help to interpret results and should consult qualified healthcare providers before making health decisions. (myadlm.org) The regulatory picture is narrower than many people assume. The Food and Drug Administration says direct-to-consumer tests are in vitro diagnostics sold without a healthcare provider, but approval of a test is not the same thing as proof that every layer of advice built on top of the result is clinically meaningful. (fda.gov) Some companies are already trying to make the service feel more medical by adding human review. Whoop says its Advanced Labs product offers 65 biomarkers, prices starting at $150 per test, and clinician-reviewed results with next steps tied to the user’s wearable data. (whoop.com) That extra review may reduce some of the risk, but it also changes the product from “instant artificial intelligence answer machine” into something closer to a digital front end for ordinary clinical interpretation. Mashable’s reporting says the core unresolved question is whether these paid services are actually better than asking a free chatbot for a plain-English explanation or waiting for a doctor who knows your case. (mashable.com) The most defensible use for these tools right now is narrow: translation first, decisions second. Even the skeptical experts in Mashable’s piece say artificial intelligence can help explain jargon, but they do not yet see strong evidence that it can reliably turn a blood panel into personalized medical advice that is accurate, actionable, and worth the subscription price. (mashable.com)