A persistent calf cramp, initially dismissed as a muscle spasm, might have proved fatal had the author not deployed a bespokebespoke/bɪˈspoʊk/L3定制的；专门设计的custom-made or tailored to meet specific requirements or preferences AI health tool trained on his medical records. For five days the pain intensified, accompanied by swelling and tenderness, yet the chiropractor treated it as a muscular issue. The AI system, however, flagged deep vein thrombosis (DVT) and directed him toward the critical diagnostic step—an ultrasound—thereby averting a potential pulmonary embolismpulmonary embolism/ˈpʌlməˌnɛri ˈɛmbəˌlɪzəm/L3肺栓塞a life-threatening condition in which a blood clot blocks an artery in the lung. This sequence illustrates how patient advocacy, augmentedaugmented/ɔːɡˈmɛntɪd/L3增强的；扩大的enhanced or supplemented by adding something to improve capability or effectiveness by machine intelligence, can intercept a trajectory that might otherwise culminate in death.
DVT, a blood clot lodged in a deep vein—typically in the leg—carries the risk of embolisation to the lungs, where it can cause sudden death. The Centers for Disease Control and Prevention characterises DVT and pulmonary embolism as serious, frequently underdiagnosedunderdiagnosed/ˌʌndərˈdaɪəɡnoʊzd/L3诊断不足的；未被充分诊断的identified or diagnosed less frequently than the actual prevalence of a condition warrants conditions; the National Heart, Lung, and Blood Institute warns that large or multiple clots can prove lethal. Despite such dangers, the healthcare system often impedes timely diagnosis: urgent care cannot perform the requisite ultrasound, and primary care offices may delay referral. Consequently, the author bypassed these fragmented pathways and proceeded directly to the emergency room, where an ultrasound revealed four clots in his left leg.
A recent study published in Science, led by researchers affiliated with Harvard Medical School and Beth Israel Deaconess Medical Center, tested a large language model on clinical reasoning tasks drawn from real emergency department cases. The model outperformed physicians in including the correct diagnosis among differentials, suggesting that doctors and AI may be safer together than either is alone. That said, a Guardian report found that one in seven people in the UK now consult AI chatbots for medical guidance instead of seeing a general practitioner—a trend that should alarm clinicians and regulators alike, given that chatbots cannot examine a patient or assume clinical responsibility.
Notwithstanding the promise of AI, its integration into healthcare demands rigorous regulation, transparency, and clinical supervision—and a concomitantconcomitant/kənˈkɑːmɪtənt/L3相伴的；伴随的naturally accompanying or associated with something, often as a secondary effect humility from institutions that too often expect patients to navigate fragmented systems unaided. The author’s experience does not argue for replacing doctors with machines; rather, it underscores that patient advocacy now possesses a new tool: a safe, personalised AI assistant that can organise records, surface urgent possibilities, and prompt the correct diagnostic step. Medicine has long relied on second opinions; the next may emanate from software, and the urgent task is to ensure that software is accurate, accountable, and deployed to save lives.
