Historians’ Methods Could Rescue Medicine from AI’s One-Dimensional Patient Summaries

A nurse and former historian at Johns Hopkins has ignited an unusual debate in internal medicine: the argument that clinicians should be trained to think like historians. In a perspective article published in the Journal of General Internal Medicine, Matthew E. Franco, who holds a doctorate in history and now works in the medical intensive care unit at Johns Hopkins Bayview Medical Center, contends that the methodologies historians have refined over centuries—deep reading, archival research, and critical textual analysis—are not merely cultural ornaments for medical education. They are practical diagnostic tools, capable of recovering patient information that conventional interviews, electronic health records, and increasingly AI-generated clinical summaries systematically destroy. The claim lands at a moment when medical schools are already wrestling with how, or whether, to teach the history of medicine, and when artificial intelligence is compressing patient lives into tidy data snapshots.

Franco’s central criticism targets the way history of medicine has typically entered the curriculum. Survey courses tend to emphasize paradigm-shifting discoveries and archetypal “great persons,” which, he argues, informs trainees about their profession’s development without giving them any new skills to apply at the bedside. The contrast he draws is with the social history—the patient interview about living conditions, occupation, and habits—which is well established in clinical training but conventionally limited to direct interviewing technique. Historians, by contrast, are trained to work with many kinds of sources, oral and written, and to extract truths that direct questioning does not readily expose. That distinction matters, Franco suggests, because the clinical interview itself is under siege. Citing a 2024 paper by Fava and colleagues, he notes that interviewing as a basic skill is threatened by skill erosion driven by the allure of technology, time pressures, shortcuts such as questionnaires and chatbots, and the hegemony of the biomedical model of disease.

The technical core of the argument rests on two methodologies borrowed from historical scholarship. The first is the microhistory approach associated with the medievalist Natalie Zemon Davis: following individual people through their archival traces, gathering fragments of lived experience from scattered documents to construct a fuller picture of a life. The second is “reading against the grain,” a technique described by the historian Marisa Fuentes in her work on the archives of slavery, in which documents created to support systematic oppression are read to “eke out extinguished and invisible but no less historically important lives.” Franco applied both in his own doctoral research on eighteenth-century lives under the Spanish empire, scouring bureaucratic databases—court cases, tax bills, military service records, property documents—and connecting the fragments to reconstruct social context. His proposal is that these same techniques give clinicians a systematic way to fill gaps in patient histories, particularly for marginalized patients who arrive with incomplete illness narratives.

The COVID-19 pandemic supplied the field test. When universities shuttered, Franco moved into pandemic response work, and from 2021 served as an operations manager for community health projects delivering vaccines to marginalized populations. The historian’s toolkit proved immediately applicable: tracking down information missing from medical records and reconciling conflicting documentation. It was in that setting that he encountered the patient he calls Ms. H, during the Omicron surge of January 2022. Hesitant and reluctant, the young woman said she needed vaccination “for immigration.” Working in clinics serving low English proficiency Latinx patients, Franco recognized a familiar pattern: the US government required proof of COVID vaccination, yet many patients feared that presenting for care without proof of legal residency would lead to deportation, unsettled by prior encounters with Customs and Border Protection.

What followed illustrates the method in action. Franco asked whether Ms. H had already received any vaccine, knowing that CBP sometimes administered a single dose before releasing unaccompanied minors from detention. She had received one dose at home, she said, but had lost all her papers crossing the river—”I lost everything.” As the two of them organized the slips of documentation she still carried, from medical encounters after crossing the Darién Gap and from her dealings with CBP, and identified what the lost papers had contained, Ms. H began voluntarily disclosing details of her transit. Then came the question that reframed the encounter: was it safe to be vaccinated while pregnant? Before Franco could explain maternal vaccination, she began describing trauma endured during her journey. Correlating her account against the fragmentary paperwork revealed a story that no structured intake form had captured.

Franco’s point is not sentimental but epistemological. No data-driven analysis, he argues, would have illuminated the violence Ms. H experienced; examining immigration rates, community vaccination uptake, or maternal immunity statistics could not have surfaced her narrative. He had initially framed her need for vaccination economically, as a condition of employment, then legally, as a condition of remaining in the country—neither approach exposed her most pressing health concern. The physiological and psychological stresses she endured, including sexual violence, have cumulative effects on health, a phenomenon the demographer Arline Geronimus has termed “weathering.” Yet such stresses can remain hidden from clinicians because of the fragmentary nature of the medical record narrative. Methodologies developed by historians of slavery, Franco argues, recover meaning from precisely this kind of fragmentary record, and they equip clinicians to identify and combat the structural violence embedded in patients’ lived experiences.

The argument acquires particular urgency from the rise of AI tools that generate patient histories by extracting information from the medical record. Such summaries, Franco warns, reduce patients to one-dimensional snapshots, and the risk of omission is heightened when an algorithm assembles the chart. A clinician trained in historical methods can see beyond the snapshot and reconstruct a fuller picture. The documentary record, in Ms. H’s case, worked to erase her experience through omission and concealment; an AI summary built on that same record would inherit its silences. This is a technical critique, not a Luddite one: the problem is not that records are digitized, but that digitization preserves and propagates whatever gaps and errors the underlying data contain.

Indeed, much of Franco’s clinical record-keeping work is strikingly mundane compared with eighteenth-century archives. Incomplete medical records typically stem from electronic systems that have never been reconciled to one another—a misspelled name, an outdated address, an unrecorded emergency department visit. Each error splinters the personal history a little further. In his eighteenth-century research, recombining fragments meant traveling between the modern sites of the early Spanish Atlantic world to triangulate the movement of people through entangled empires and marketplaces. In the clinic, a few clicks updating demographic information can bring the whole patient into focus, and unlike long-dead archival subjects, living patients can say where the record is wrong. Still, not everything rectifies quickly: many Spanish-speaking patients brought paper vaccination records from their countries of origin, manually entered into electronic systems and preserved indefinitely, errors and all.

Whether medical schools will institutionalize any of this remains open. Surveys of North American medical schools by Kent and Ward found that the history of medicine is most often non-mandatory, delivered as lectures from clinical faculty, while humanities scholarship reaches trainees mainly through elective offerings and reading lists such as those promoted by the Gold Foundation. Franco’s proposal is more demanding: not reading history, but training in historical method—deep reading, archival reasoning, textual criticism—so that clinicians learn to view the patient interview as a dialogue that constructs a narrative, noting gaps the way historians see absences in an archival record, and chasing those gaps into new lines of questioning. Many experienced clinicians already work this way intuitively; the proposal is to teach it deliberately rather than leaving it to be discovered through experience.

The outcome for Ms. H suggests the stakes. The questions Franco asked revealed what he calls an unspeakable horror, but they also allowed him to meet her actual needs: he scheduled her an intake appointment the following week at mental health and maternal health clinics specializing in trauma-informed care. Her vaccination record satisfied CBP; she remained in the state and delivered her baby within the health system months later. Franco, who now applies the same record-reconstruction skills in the intensive care unit—searching for missing vaccination records, prior primary care notes, or evidence of emergency department visits—concedes that how such methods would be taught to clinical trainees is beyond the scope of his paper. But his conclusion is pointed: “reading the silences” in a patient’s record can change what happens to that patient, and a profession racing to automate its documentation may need historians more than it thinks.

Subject of Research: Using historical research methodologies such as deep reading and archival analysis to improve clinical practice and patient history-taking

Article Title: Historical Thinking in the Clinic

Article References: Franco, M. E. (2026). Historical Thinking in the Clinic. Journal of General Internal Medicine. https://doi.org/10.1007/s11606-026-10759-y

Image Credits: AI Generated

DOI: 10.1007/s11606-026-10759-y

Keywords: medical humanities, history of medicine, clinical education, AI in healthcare, electronic health records, patient interviews, social determinants of health, structural violence, microhistory, weathering, community health, Journal of General Internal Medicine