Weekly AI Watch

Every week, what changed and what it means for you. No hurry, no alarm.

Week of 20 November 2026 | Past editions15,959 papers found · 15,002 read · 5,264 belonged · 42 stories passed every check

Observatorio de la
Inteligencia Artificial

Cada semana, qué cambió y qué significa para usted. Sin prisa y sin alarma.

Semana del 20 de noviembre de 2026 | Ediciones anteriores15,959 estudios encontrados · 15,002 leídos · 5,264 correspondían · 42 notas pasaron todas las comprobaciones

weeklyAI · Week of 20 November 2026weeklyAI · Semana del 20 de noviembre de 2026

← The good news← Las buenas noticias

experiment · Oral and maxillofacial surgery · la publicación, 26 sep 2026 · gratis

Un programa reescribió informes médicos en palabras llanas, y la gente entendió más

En un estudio alemán, quienes leyeron la versión simplificada acertaron más preguntas sobre su contenido, aunque fue en un solo hospital y en alemán.

Versión breve · la versión detallada sigue, unos 5 min

Pregunte al chat de weeklyAI

Pregúnteme sobre la IA en su vida: le señalo las notas que weeklyAI ha publicado sobre el tema, esta semana y antes.

Las conversaciones se guardan mientras exista weeklyAI, para mejorar la publicación. Se responde en el idioma en que usted escribe.

El estudio, de un vistazo
Quiénes
Personas sin formación médica que leyeron un informe médico
Cuántos
576 participantes
Dónde
Alemania, Hospital Universitario de Maguncia
Cuándo
No lo dice el texto
Tipo de estudio
experiment
Quién lo hizo
Hospital Universitario de Maguncia, Alemania
El límite que importa
Todo se hizo en alemán, en un solo hospital, y sin seguimiento posterior.
Porcentaje de respuestas correctas sobre el contenido del informe, versión simplificada frente a original
Reporte quirúrgico, simplificado92.9%
Reporte quirúrgico, original75.2%
Resumen de alta, simplificado88.7%
Resumen de alta, original61.8%
Tomografía, simplificada87.9%
Tomografía, original68.8%
Resonancia magnética, simplificada80.3%
Resonancia magnética, original53.7%

Son los aciertos en preguntas de opción múltiple sobre cinco informes reales; el de patología no aparece porque su mejora no fue concluyente, y todo se hizo en alemán en un solo hospital.

Informe original frente a informe reescrito en palabras llanas

Versión simplificadafrente aVersión original

Más respuestas correctas en cuatro de los cinco documentos

Versión simplificadafrente aVersión original

Mejor calificada en claridad, estructura y utilidad

Versión simplificadafrente aVersión original

Más fácil de leer según dos escalas de lectura

Informe de patología simplificadofrente aInforme de patología original

Mejoró, pero la diferencia pudo deberse al azar

La pregunta útil en su clínica es si ofrecen una versión en lenguaje sencillo junto al informe original.
Lectura de weeklyAI

Usted sale de la clínica con una hoja impresa. Tiene siglas, palabras en latín y frases que parecen escritas para otro médico. Con ese papel debe entender qué le encontraron y qué sigue.

Un grupo de investigadores del Hospital Universitario de Maguncia, en Alemania, tomó cinco informes reales: uno de resonancia magnética, uno de tomografía, uno de cirugía, uno de patología y una hoja de alta. Le pidieron a ChatGPT 4.0, un programa que escribe solo, que reescribiera cada uno para un lector de unos 11 o 12 años. La orden fue clara: conservar toda la información clínica.

Después repartieron esos informes entre 576 personas sin estudios de medicina. Cada una leyó una sola versión, la original o la simplificada, y respondió un cuestionario y cinco preguntas de opción múltiple sobre lo que había leído.

Las respuestas correctas subieron del 62.2% al 82.2% en conjunto. En el informe de resonancia, pasaron del 53.7% al 80.3%. La gente también calificó las versiones llanas como más claras y más útiles. Los propios autores repitieron la simplificación diez veces por cada informe y compararon cada versión con una lista de datos del original. No faltó ningún dato médico.

Ahora, lo que el estudio no muestra. Todo se hizo en alemán, en un solo departamento de cirugía oral y maxilofacial, y no participaron personas con formación médica ni menores de 18 años. Nadie comprobó cada texto con un especialista independiente, y a los participantes no se les siguió después. Entender mejor una prueba no es lo mismo que recibir mejor atención.

El informe de patología, el más técnico, mejoró menos: 61.2% frente a 51.3%. Esa diferencia pudo deberse al azar, así que los documentos más densos pueden seguir siendo difíciles.

El programa que hizo esto fue ChatGPT 4.0. Los investigadores usaron los mismos ajustes en todos los informes y cambiaron solo un poco la orden que le daban, según el tipo de documento. La pregunta útil en su clínica es si ofrecen una versión en lenguaje sencillo junto al informe original. Si existe, pídala; si no, pregunte si puede. Y guarde siempre el original.

Qué significa para usted

En su próxima consulta, revise si el informe que le entregan viene acompañado de una versión en palabras sencillas. En el estudio, esa versión se entendió mejor, aunque fue en alemán y en un solo hospital. Si no la ve, pregunte si pueden dársela, y conserve siempre el original.

Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4

Quién pagó: Los autores declaran que no recibieron fondos, subvenciones ni otro tipo de apoyo durante la preparación del manuscrito y que no tienen conflictos de interés; se utilizó ChatGPT 4.0 (OpenAI) como herramienta de simplificación y ChatGPT 5.0 para el refinamiento del lenguaje y la corrección de pruebas.

No tome esto como consejo médico profesional.

Versión detalladaLos pasajes copiados del artículo, las ilustraciones y cada fuente con cuánto leímos de ella · 1087 palabras · unos 5 minLeerla →Cerrar

Un chatbot reescribió informes médicos en alemán. Los pacientes entendieron más, pero solo en un hospital

El estudio se hizo con 576 personas sin formación médica, en una sola clínica de Alemania. Lo que muestra, y lo que deja sin responder.

Los informes médicos se escriben con abreviaturas y términos de anatomía, y el paciente los lee sin saber bien qué le encontraron ni qué sigue. El estudio que aquí se cuenta partió de ese problema: los informes médicos se escriben para otros médicos, con abreviaturas y términos de anatomía, y para la mayoría de la gente eso es una barrera para entender su estado de salud y su tratamiento1. La investigación la firmaron Stephan, Schumacher, Al-Nawas, Kämmerer y Thiem, del Departamento de Cirugía Oral y Maxilofacial del Hospital Universitario de Maguncia, en Alemania, y no recibió financiamiento. Los autores declaran no tener conflictos de interés.

Piense en la hoja que le entregan al salir de urgencias. El artículo describe un estudio en el que buena parte de los pacientes sale sin entender bien su diagnóstico, su tratamiento o las indicaciones de seguimiento, y muchos no recuerdan los síntomas de alarma aunque la información se les haya dado de palabra o por escrito2. El artículo describe un estudio en el que los folletos de oncología y cardiología suelen estar escritos a nivel universitario cuando las autoridades de salud pública recomiendan un nivel de sexto grado para materiales dirigidos a pacientes3. Lo curioso es que el problema no es la falta de información: es que la información está escrita para otro lector.

Lo que probaron fue sencillo de describir. Tomaron cinco documentos médicos reales y anonimizados: un informe de resonancia magnética, uno de tomografía, un reporte quirúrgico, uno de patología y un resumen de alta. Cada uno se reescribió con ChatGPT 4.0 mediante una instrucción estándar que pedía conservar toda la información clínica, no inventar nada y mantener la coherencia del original, en un lenguaje apto para una persona con un nivel de lectura de unos 11 o 12 años4. Se repitió el ejercicio diez veces por documento para revisar que no se perdiera nada.

Después, 576 personas sin conocimientos médicos leyeron una sola versión de un solo documento, asignada al azar: 289 leyeron la simplificada y 287 la original5. Contestaron un cuestionario de opinión y cinco preguntas de opción múltiple sobre el contenido. En los índices de facilidad de lectura, las versiones simplificadas salieron mejor: 48.4 frente a 22.9 en la escala Flesch, donde más alto es más fácil, y 48.3 frente a 59.2 en el índice LIX, donde más bajo es más simple6. En las calificaciones de claridad, estructura y utilidad, los textos simplificados ganaron en las tres categorías7.

En las preguntas de comprensión, la diferencia se vio en cuatro de los cinco documentos. Los reportes quirúrgicos simplificados dieron 92.9 por ciento de respuestas correctas frente a 75.2; los resúmenes de alta, 88.7 frente a 61.8; las tomografías, 87.9 frente a 68.8; las resonancias, 80.3 frente a 53.78. Y algo importante para quien teme que un chatbot "invente": al revisar las diez versiones de cada texto, no se encontró pérdida de información clínica9.

Aquí vienen los límites, y son grandes. Todo se hizo en alemán, en un solo entorno lingüístico y cultural, como advierten los propios autores10. Los participantes se reclutaron en un solo departamento clínico, lo que puede sesgar la muestra11. No se recabó el nivel educativo de nadie, por el diseño anónimo, y ese dato importa porque la escolaridad va de la mano con la comprensión de temas de salud12. En los informes de patología, que son los más técnicos, la mejora existió (61.2 frente a 51.3 por ciento) pero no alcanzó significancia estadística13. Ningún experto médico independiente revisó cada salida del sistema, y a nadie se le dio seguimiento en el tiempo.

Hay otros trabajos que ayudan a dimensionar esto. La gente ya está usando estas herramientas para descifrar sus papeles.

Así lo lee weeklyAI. Lo que este estudio mide es una cosa muy concreta: si el mismo documento, reescrito, deja a una persona común con más respuestas correctas en un cuestionario. No mide si esa persona se cuidó mejor, si tomó mejores decisiones ni si recordó algo una semana después. Lo que sí sugiere es que el problema no es solo de vocabulario: la forma en que se organiza un texto también pesa. Un informe puede tener todas las palabras correctas y aun así no dejar claro qué sigue.

El artículo describe un estudio en el que se señala que estos sistemas pueden producir afirmaciones que suenan bien pero son falsas, y que pueden arrastrar sesgos. En este estudio alemán, en cambio, no se encontró pérdida de información clínica.

Esta lectura se equivocaría si, al probarse en otros idiomas, las respuestas correctas no subieran, o si subieran igual en patología que en los demás documentos, o si el texto gustara más sin mejorar la comprensión. También se equivocaría si el beneficio desapareciera al repetir la prueba semanas después, cosa que aquí nadie hizo.

Lo que weeklyAI espera, y esto es una expectativa, no un hallazgo: que el beneficio sea mayor en los documentos con más datos duros —imágenes, altas— que en los muy especializados, como patología, donde el propio estudio ya vio que la mejora no fue concluyente. Y espera que el efecto dependa de algo que aquí no se midió: cuánta escuela trae cada lector. Si tiene un informe en la mano, puede pedir en su clínica una versión en lenguaje sencillo y comprobar si de verdad le aclara el diagnóstico y los siguientes pasos. Si no existe, puede preguntar si se la pueden preparar. Y al recibirla, conviene contrastarla con el original y anotar las dudas para llevarlas a la consulta.

Los autores lo dicen sin rodeos: el médico sigue siendo responsable de la exactitud y la pertinencia de lo que se le informa al paciente, y estas herramientas deben verse como un apoyo, no como un reemplazo de la conversación humana16. Piden además que se investigue en varios idiomas y contextos culturales17, que es justo lo que aquí no se hizo.

El informe con la palabra "hiperintensa" sigue en su carpeta. Lo que este estudio muestra es que una segunda versión, en palabras llanas, puede hacer que usted conteste mejor qué le encontraron. Lo que no muestra es si eso lo va a cuidar mejor. La pregunta que puede llevar a su próxima cita es sencilla: ¿me puede dar esto en palabras que entienda, y me lo puede explicar?

De dónde sale cada dato de contexto, y cuánto leímos de cada documento

  1. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - el artículo del que trata esta nota — el artículo completo — el pasaje: "However, medical reports are often written in highly specialized language, characterized by abbreviations, anatomical terminology, and discipline-specific phrasing. For most patients, this complexity poses a major barrier to understanding their health status, treatment options, and recommendations."
  2. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Research consistently demonstrates that a substantial proportion of patients leave emergency departments without fully understanding their diagnosis, treatment, or follow-up instructions [ 8 , 9 ]. Many patients fail to recall critical warning symptoms or adhere to discharge recommendations, regardless of whether the information is provided verbally or in written form."
  3. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - el artículo del que trata esta nota — el artículo completo — el pasaje: "For example, educational resources in oncology and cardiology are commonly written at a university level, despite public health authorities recommending a sixth-grade reading level for patient materials [ 34 ]."
  4. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - el artículo del que trata esta nota — el artículo completo — el pasaje: "For each document, a context-specific prompt instructed ChatGPT to rephrase the text into accessible language suitable for laypersons with a reading level of approximately 11–12 years. The instructions explicitly required preservation of all clinically relevant information, exclusion of fabricated content, and maintenance of the original logical and structural coherence."
  5. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - el artículo del que trata esta nota — el artículo completo — el pasaje: "A total of 576 participants completed the survey, with 289 evaluating the simplified and 287 the original text versions, as summarized in Table 1 ."
  6. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Simplified reports achieved significantly higher readability (FRE 48.4 ± 5.0 vs. 22.9 ± 5.2; p < 0.0001; LIX 48.3 ± 3.2 vs. 59.2 ± 2.5; p = 0.004)."
  7. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Across all document types, patients rated simplified texts significantly higher in clarity, structure, and usefulness ( p < 0.001), with significantly improved comprehension accuracy (e.g., MRI 80.3% vs. 53.7%; p < 0.001). No loss of medical information was observed."
  8. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Participants demonstrated significantly higher comprehension accuracy for simplified surgical reports (92.9 % vs. 75.2 %; p < 0.001; n = 184 / 157), discharge summaries (88.7 % vs. 61.8 %; p < 0.001; n = 196 / 158), CT reports (87.9 % vs. 68.8 %; p < 0.001; n = 157 / 186), and MRI reports (80.3 % vs. 53.7 %; p < 0.001; n = 188 / 164)."
  9. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - el artículo del que trata esta nota — el artículo completo — el pasaje: "For all text types, each of the ten repetitively simplified versions retained the full medical content of the corresponding original report. The simplification process, performed repeatedly for validation, did not result in any loss of relevant information."
  10. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - el artículo del que trata esta nota — el artículo completo — el pasaje: "First, the study was conducted in German within a single linguistic and cultural setting, which may limit the generalizability of the findings to other languages and healthcare systems."
  11. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Furthermore, participants were recruited from a single clinical department, which may introduce selection bias and limit the representativeness of the study population."
  12. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - el artículo del que trata esta nota — el artículo completo — el pasaje: "In addition, information on participants’ educational background was not collected due to the fully anonymous study design. As educational level is closely associated with health literacy, future studies should incorporate this variable to allow a more detailed interpretation of comprehension outcomes."
  13. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - el artículo del que trata esta nota — el artículo completo — el pasaje: "For pathology reports, comprehension rates were higher for the simplified version (61.2 % vs. 51.3 %; n = 168 / 136), although this difference was not significant (p = 0.0669)."
  14. Kobayashi K, Wu Z, Tomoi H, Masuda K, Larson HJ, Leung K, et al. (2026). AI Chatbot Use for Human Papillomavirus Vaccine Literacy in Japan. JAMA Network Open. 10.1001/jamanetworkopen.2026.34696 — solo el resumen - no se pudo obtener el texto completo — el pasaje: "In this RCT of female caregivers in Japan, an AI-driven HPV vaccine chatbot produced higher adjusted HPV vaccine literacy than a government leaflet immediately after intervention and at 2-week follow-up, without a measurable short-term effect on vaccination decision."
  15. Wei Q, Xiao L, Zhang X, Wei H, Wang R, Ou S, et al. (2026). Real-World Use and Expert-Guided Optimization of a Large Language Model–Based Pediatric Health Assistant: Retrospective Observational Study. Journal of Medical Internet Research. 10.2196/105346 — solo el resumen - el texto completo está tras una suscripción — el pasaje: "The leading reasons for consultation were symptom consultation (7845/16,074, 48.8%), medical report interpretation (2531/16,074, 15.7%), and triage guidance (2515/16,074, 15.6%); among symptom consultation sessions, fever (2719/16,074, 34.7%) and cough (2349/16,074, 29.9%) were most common."
  16. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Physicians remain ultimately responsible for the accuracy and appropriateness of the information provided to patients, underscoring the need for institutional oversight and transparent review mechanisms. AI tools such as ChatGPT should thus be regarded as complementary aids rather than replacements for human communication."
  17. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Future research should investigate AI-based text simplification in multilingual settings to assess its generalizability across different linguistic and cultural contexts."

Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4

Quién pagó: Los autores declaran que no recibieron fondos, subvenciones ni otro tipo de apoyo durante la preparación del manuscrito y que no tienen conflictos de interés; se utilizó ChatGPT 4.0 (OpenAI) como herramienta de simplificación y ChatGPT 5.0 para el refinamiento del lenguaje y la corrección de pruebas.

No tome esto como consejo médico profesional.

Los hallazgos de otros estudios que aquí se mencionan provienen de este documento; esos estudios no se consultaron uno por uno.

experiment · Oral and maxillofacial surgery · the paper, 26 Sep 2026 · free

A Chatbot Made Medical Reports Easier to Answer. The Test Was in German, in One Clinic.

In a study of 576 people with no medical training, plain-language rewrites of five report types raised test scores. No doctor checked every rewrite, and no one followed the patients afterward.

Short version · the longer version follows, about 6 min

Ask the weeklyAI chat

Ask me about AI in your life: I point you to the stories weeklyAI has published on it, this week and before.

Conversations are saved for as long as weeklyAI exists, to improve the publication. Answers come in the language you write in.

The study at a glance
Who
People with no medical training reading five kinds of medical report
How many
576 participants
Where
Germany (University Medical Center Mainz)
When
not stated in the passages
Kind of study
experiment
Who did it
University Medical Center Mainz, Germany
The limit that matters
It ran in German, in one hospital department, and no independent medical expert checked every rewrite.
Share of questions answered correctly, original report vs chatbot rewrite
MRI report, original53.7%
MRI report, rewrite80.3%
CT report, original68.8%
CT report, rewrite87.9%
Surgical report, original75.2%
Surgical report, rewrite92.9%
Discharge summary, original61.8%
Discharge summary, rewrite88.7%
Pathology report, original51.3%
Pathology report, rewrite61.2%

Percent of multiple-choice questions answered correctly after reading each report version; the pathology gap was not shown to be real rather than chance, and all texts were in German.

Original medical reports vs chatbot rewrites

Original reportagainstChatbot rewrite

Readers answered more questions correctly after the rewrite for MRI, CT, surgical and discharge reports

Original pathology reportagainstRewritten pathology report

More correct answers after the rewrite, but the gap could not be shown to be real rather than chance

Original reportagainstChatbot rewrite

Readers rated the rewrites higher for clarity, structure and usefulness

Original reportagainstChatbot rewrite

Ten repeated rewrites of each report kept all the medical content

So the study shows that a test of understanding improved. It does not show that patients took their medicine, kept their follow-up appointments or made better decisions.
weeklyAI's reading

You leave a clinic with a printed report, and a word on it could come from a Latin textbook. You are expected to decide something with that paper in your hand.

Researchers at the University Medical Center Mainz, in Germany, took five real reports of the kinds patients often get: an MRI report, a CT report, a surgical report, a pathology report and a discharge summary. They had ChatGPT 4.0 rewrite each one for a reader of about 11 or 12 years. The instructions told the chatbot to keep every medical detail and to invent nothing.

They then gave 576 people one version of one report, original or rewritten, and asked five multiple-choice questions about it. Nobody in the group had medical training; people with a health care background and anyone under 18 were excluded. The assignment was random, and readers did not know which version they held.

The plain versions won. Overall, 82.2 percent of the answers were correct, against 62.2 percent for the originals. For the MRI report, correct answers rose from 53.7 percent to 80.3 percent. People also rated the rewrites clearer and more useful, and the researchers found that ten repeated rewrites of each report lost no medical information.

One kind of report held out. For pathology reports, correct answers went from 51.3 percent to 61.2 percent, and the researchers could not show that this gap was real rather than chance. The most technical reports may stay hard even after rewriting.

Grant the study its careful design. It gave readers no time limit, and it used the same questions for both versions. It also asked people to rate how clear the text was, alongside the test of understanding.

The study was carried out in German, in one department of one German hospital. Whether the same gain would appear in Spanish, Portuguese or English, or in the clinic where you are treated, was not tested. No independent medical expert checked every rewritten report, and nobody followed the readers home.

So the study shows that a test of understanding improved. It does not show that patients took their medicine, kept their follow-up appointments or made better decisions.

Look for a clinic or health service that offers a plain-language version of a report next to the original, and keep the original in hand.

What this means for you

What the study measured was a test of reading, and the plain versions helped people answer questions about a report they had just read. It did not follow anyone home, so it cannot say whether anyone took a medicine or made an appointment because of the rewrite. When a clinic prints a report in words you can follow, the useful thing to ask is whether both versions are there to compare.

Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4

Who paid: The authors state that no funds, grants or other support were received during the preparation of the manuscript, and they declare no competing interests; ChatGPT 4.0 (OpenAI) was used as the simplification tool and ChatGPT 5.0 for language refinement and proofreading.

Do not take this as professional medical advice.

The longer versionThe passages copied from the paper, the pictures, and every source with how much of it we read · 1165 words · about 6 minRead it →Close

# A Chatbot Rewrote Five Medical Reports. Readers Scored Higher on What They Understood.

The test was in German, in one hospital department, with five documents. It says nothing yet about your clinic, your language, or your care.

A pathology report arrives in the mail. It says "adenocarcinoma, moderately differentiated, margins clear." Whether that sentence frightens a family or reassures them depends on a translation nobody performed.

That translation, in this study, was done five times: once each for an MRI report, a CT report, a surgical report, a pathology report and a discharge summary. Five anonymized German documents, written by physicians, were handed to ChatGPT 4.0 (OpenAI, San Francisco, CA, USA). The prompt asked it to rewrite each into language a person with roughly 11 to 12 years of reading could follow, to keep every clinically relevant fact, to add nothing invented and to preserve the report's order and logic1. Each document was simplified ten separate times, in fresh sessions, and a checklist built from the original was used to compare every output against it1. For every document type, each of the ten simplified versions kept the full medical content of the original, the authors report2.

Then 576 people with no medical training read one report, either the original or the rewrite, and answered questions. The split was 289 to 2873. The authors are at the Department of Oral and Maxillofacial Surgery and Facial Plastic Surgery, University Medical Center Mainz, Germany. No outside funding is declared, and the authors declare no competing interests.

On two standard readability scores, the rewrites won. The Flesch Reading Ease score, where higher means easier, rose from 22.9 to 48.4. The LIX index, where lower means simpler, fell from 59.2 to 48.34. Readers rated the rewrites higher on clarity, structure and usefulness, and answered more comprehension questions correctly: 82.2 percent against 62.2 percent overall5. On the MRI report, 80.3 percent of answers were right after the rewrite, against 53.7 percent after the original6. Surgical reports, discharge summaries and CT reports showed the same pattern6.

One document resisted. On the pathology report, correct answers went from 51.3 percent to 61.2 percent, and that gap was not statistically significant7. The authors suggest why in their discussion: pathology carries the densest terminology and the most abstract classifications, the material least willing to become plain.

The limits start with language and place. The whole study ran in German, in one linguistic and cultural setting, which the authors say may limit how far the results travel8. All participants came from a single clinical department, which may skew who volunteered9. Education level was never recorded, because the survey was fully anonymous, so the gains cannot be separated from schooling10. Comprehension was tested with multiple-choice questions, and the per-question figures cover four questions per text type. No independent medical expert checked every simplified output. And the authors are explicit that ChatGPT's rewrites can vary with model updates, prompt wording and settings.

Consider what the clinic visit covers. You leave with a sheet of paper, and the paper was written for the doctor who comes after your doctor: abbreviations, anatomical terms, a phrase like "discipline-specific" that no one at your kitchen table uses11. This is not a failure of any one physician. It is what documentation is for.

The gap has been measured elsewhere. Research cited in this study finds that a substantial share of patients leave emergency departments without fully understanding their diagnosis, treatment or follow-up, and that many cannot recall warning symptoms even when the information was given both aloud and in writing12. Patient materials in oncology and cardiology are commonly written at university level, while public health authorities recommend a sixth-grade reading level13. The mismatch is old and wide.

The mechanism is a rewrite, not a summary. The prompt did not ask ChatGPT to shorten; it asked for the same content in easier words, with invented content forbidden and the original structure kept1. That distinction matters to anyone who has received a "patient-friendly" leaflet that dropped the finding they most needed.

This is not the first attempt from this group. An earlier study of theirs found that ChatGPT-simplified radiology reports improved readers' impressions of clarity, empathy and readability14. The new work extends the test to five document types and, more importantly, adds questions with right and wrong answers. A report that feels clearer is one thing; a reader who can answer correctly about it is another.

The number worth picturing: roughly four in five questions answered correctly after a rewrite, against about three in five after the original5. Not perfection, and not a small difference either. A reader who fails two of five questions about their own scan is a reader who may misremember the follow-up date.

What follows next, if this holds, is a workflow. A clinic generates the professional report for the file and a second, plain version for the patient, from the same source, without extra work for the physician. The authors themselves call for testing in multiple languages and health systems before that is treated as settled15. They also hold a line: the physician remains responsible for accuracy and appropriateness, and the tool is an aid, not a replacement for human communication16.

Here is weeklyAI's reading. The pattern is deference: people trust a document because of who issued it, not because they checked it. A plain-language rewrite arriving from the same hospital carries the same authority as the original, and possibly more, because it is easier to accept. That is the trade: you gain comprehension and you lose the friction that might have sent you back to the original with a question. This reading would be wrong if readers told the rewrite came from a neighbor rated it just as clear and answered just as many questions right. Until someone tests that, keep both documents side by side and compare the findings, the dates and the next steps yourself.

A second pattern worth naming: specialist language drifts out of reach as a field matures, and no amount of good writing by individual doctors reverses it. Expect to need a plain version every time a new report arrives, not once. Ask your clinic whether a plain-language version exists as routine, who produced it, and whether a clinician reviewed it before it reached you. A plain version without that review is a guess in comfortable clothing.

What this study makes possible for you is narrow and real. The next time a report comes back unreadable, there is now evidence, from one German hospital, that a chatbot asked to preserve every fact can produce a version ordinary readers understand better. Bring the plain version to your appointment, with the original in the other hand, and ask the clinician to confirm that the two say the same thing.

The pathology report in the mail may still be the hardest one to translate. But the question to ask has changed: not "what does this mean," but "is there a plain version, and who checked it?"

Where each piece of context comes from, and how much of it we read

  1. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - the article this story is about — the whole article — the passage: "For each document, a context-specific prompt instructed ChatGPT to rephrase the text into accessible language suitable for laypersons with a reading level of approximately 11–12 years. The instructions explicitly required preservation of all clinically relevant information, exclusion of fabricated content, and maintenance of the original logical and structural coherence."
  2. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - the article this story is about — the whole article — the passage: "For all text types, each of the ten repetitively simplified versions retained the full medical content of the corresponding original report. The simplification process, performed repeatedly for validation, did not result in any loss of relevant information."
  3. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - the article this story is about — the whole article — the passage: "A total of 576 participants completed the survey, with 289 evaluating the simplified and 287 the original text versions, as summarized in Table 1 ."
  4. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - the article this story is about — the whole article — the passage: "Simplified reports achieved significantly higher readability (FRE 48.4 ± 5.0 vs. 22.9 ± 5.2; p < 0.0001; LIX 48.3 ± 3.2 vs. 59.2 ± 2.5; p = 0.004)."
  5. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - the article this story is about — the whole article — the passage: "Across all document types, patients rated simplified texts significantly higher in clarity, structure, and usefulness ( p < 0.001), with significantly improved comprehension accuracy (e.g., MRI 80.3% vs. 53.7%; p < 0.001). No loss of medical information was observed."
  6. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - the article this story is about — the whole article — the passage: "Participants demonstrated significantly higher comprehension accuracy for simplified surgical reports (92.9 % vs. 75.2 %; p < 0.001; n = 184 / 157), discharge summaries (88.7 % vs. 61.8 %; p < 0.001; n = 196 / 158), CT reports (87.9 % vs. 68.8 %; p < 0.001; n = 157 / 186), and MRI reports (80.3 % vs. 53.7 %; p < 0.001; n = 188 / 164)."
  7. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - the article this story is about — the whole article — the passage: "For pathology reports, comprehension rates were higher for the simplified version (61.2 % vs. 51.3 %; n = 168 / 136), although this difference was not significant (p = 0.0669)."
  8. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - the article this story is about — the whole article — the passage: "First, the study was conducted in German within a single linguistic and cultural setting, which may limit the generalizability of the findings to other languages and healthcare systems."
  9. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - the article this story is about — the whole article — the passage: "Furthermore, participants were recruited from a single clinical department, which may introduce selection bias and limit the representativeness of the study population."
  10. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - the article this story is about — the whole article — the passage: "In addition, information on participants’ educational background was not collected due to the fully anonymous study design. As educational level is closely associated with health literacy, future studies should incorporate this variable to allow a more detailed interpretation of comprehension outcomes."
  11. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - the article this story is about — the whole article — the passage: "However, medical reports are often written in highly specialized language, characterized by abbreviations, anatomical terminology, and discipline-specific phrasing. For most patients, this complexity poses a major barrier to understanding their health status, treatment options, and recommendations."
  12. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - the article this story is about — the whole article — the passage: "Research consistently demonstrates that a substantial proportion of patients leave emergency departments without fully understanding their diagnosis, treatment, or follow-up instructions [ 8 , 9 ]. Many patients fail to recall critical warning symptoms or adhere to discharge recommendations, regardless of whether the information is provided verbally or in written form."
  13. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - the article this story is about — the whole article — the passage: "For example, educational resources in oncology and cardiology are commonly written at a university level, despite public health authorities recommending a sixth-grade reading level for patient materials [ 34 ]."
  14. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - the article this story is about — the whole article — the passage: "A previous study from our group demonstrated that ChatGPT-simplified radiology reports significantly improved patient perceptions of clarity, empathy, and readability compared with original AI-generated texts [ 21 ]."
  15. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - the article this story is about — the whole article — the passage: "Future research should investigate AI-based text simplification in multilingual settings to assess its generalizability across different linguistic and cultural contexts."
  16. Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4 - the article this story is about — the whole article — the passage: "Physicians remain ultimately responsible for the accuracy and appropriateness of the information provided to patients, underscoring the need for institutional oversight and transparent review mechanisms. AI tools such as ChatGPT should thus be regarded as complementary aids rather than replacements for human communication."

Stephan, D., Schumacher, S., Al-Nawas, B. et al. (2026). Improving patient communication and understanding of medical reports through text simplification with ChatGPT: a comparative study. Oral and Maxillofacial Surgery. https://doi.org/10.1007/s10006-026-01643-4

Who paid: The authors state that no funds, grants or other support were received during the preparation of the manuscript, and they declare no competing interests; ChatGPT 4.0 (OpenAI) was used as the simplification tool and ChatGPT 5.0 for language refinement and proofreading.

Do not take this as professional medical advice.

The findings of other studies mentioned here come from this document; those studies were not opened one by one.

The free weekly drip by email

One email a week: the free story of every section, with its picture, and the headlines of the rest.

No payment, nothing to cancel. Unsubscribe from any issue.

El goteo semanal gratis por correo

Un correo por semana: la nota gratuita de cada sección, con su imagen, y los titulares de las demás.

Sin pago y sin nada que cancelar. Puede darse de baja desde cualquier envío.

From the house

About

Every week the new research on AI is read here, and this page tells you what was found, how far it can be trusted, and what you can do with it. This page is free, and every line on it says what was found.

The studies are read by a machine built and ruled by a professor of communication at Sacramento State and Universidad Nacional Autónoma de México (UNAM). Who makes this →

Diego Bonilla, Ph.D.

Full Professor of Communication Studies, Sacramento State, California, United States
Graduate instructor, Universidad Nacional Autónoma de México (UNAM), Mexico

hypergraphia.com

I am a professor of communication at Sacramento State, where I created the course Information Management and Privacy and have taught it many times, and at UNAM, in Mexico City, I teach the graduate course Electronic and Generative Literature with AI. In both I teach AI. I have taught since 1989, at first mathematics and physics to children who found them hard, and since 1997 one subject that has changed its name many times: media literacy, computer literacy, digital literacy, information literacy, ICT literacy, and now AI literacy.

This page exists because most people cannot go back to a university, and these systems are changing their work, their children's schooling, their parents' care and their vote all the same. What a university teaches about them should reach the people who need it, every week, in plain language, without enrolling anywhere. That is what I do here, and it is what I do on Thursday mornings in a seminar for the Renaissance Society at Sacramento State, built on this publication, where anyone can send a question and hear it answered.

In 2003 my doctorate at Syracuse measured what a machine can infer about a person from the traces they leave, and what I found became the privacy course I still teach. In 2007 and 2009 I sat on the committee at ETS in Princeton that set the national standard for digital literacy, and from 2014 to 2016 I was the California State University's ICT Literacy Ambassador. Six names, one job: helping people not be at the mercy of the newest machine.

I have also worked on the other side of the counter. In 2000 I began making films that a machine assembles differently for every viewer; the first of them took first prize at the Moscow International Film Festival in 2004, two decades before the industry shipped a generative model. A generative video poem I made with a poet at UNAM is in the field's definitive anthology, and in 2025 my AI short film The Human Futures Market took Best AI Short Film at the Red Rocks AI Film Festival. My recent published research is on AI and elections, and on how these models represent the cultures they were not trained on. I do not read the papers myself, and this page will never tell you I do. I built the machine that reads them and wrote the rules it reads by, and I analyze what it gives me the way I have taught my students to analyze: asking how it knows, and who benefits from what it says.

The vendors began teaching AI literacy two years ago, because before that there was nothing to sell. I am not against them. I am simply not one of them, and no one who sells these systems pays for this page. Your subscription does, and that is what keeps it honest.

The record, dated
  • 1989 Teaching begins, Mexico City
  • 1995 Hypergraphia, an early Spanish-language web and hypermedia publisher
  • 1996 Mercadotecnia e imagen en Internet, with Jesús del Toro, a book on marketing on the Internet, among the first in Spanish, Editorial Iberoamérica
  • 1999 Autorretrato, first generative poem, published by UNAM's Instituto de Investigaciones Filológicas
  • 2003 Ph.D., Syracuse University, Doctoral Prize, for the dissertation The Medium Is the Measure of Itself
  • 2004 A Space of Time: First Prize, Internet/Multimedia, XXVI Moscow International Film Festival
  • 2007, 2009 ETS iSkills committee, Princeton
  • 2008 Making Sense of Tracking Data, a book on how the traces a person leaves online allow inferences about them, including how they think, VDM Verlag
  • 2011 Keynote on generative work, UNAM
  • 2014–2016 ICT Literacy Ambassador, California State University · California Open Educational Resources Council
  • 2019 Big Data, with Rodolfo Mata: Electronic Literature Collection, Volume 4
  • 2020 Tablada Hipertextual, with Rodolfo Mata, UNAM: the computer reads a poet's words and builds a different reading from them, the ground the language models now work on
  • 2021 Uku Pacha, a generative VR film: world premiere, Vancouver International Film Festival
  • 2023 Keynote on the representation of Latin American cultures in AI, Ibero-American Meeting of Audiovisual Narratives, Trujillo, Peru · Graduate teaching at UNAM begins
  • 2023–2026 Directed the faculty learning community Advancing AI Literacy, Sacramento State · Principles for Cognitive and Pedagogical Sovereignty
  • 2024 A weekly AI publication run on real pipelines, every week since
  • 2025 Best AI Short Film, Red Rocks AI Film Festival
  • 2026 Keynote, with Nancy Huante-Tzintzun, Latin American and Caribbean Digital Humanities symposium, UT San Antonio · AI Watch, a weekly seminar for the Renaissance Society at Sacramento State

The research record →

TertulIA: weeklyAI.watch/tertulia/en/Pictures are illustrations generated by weeklyAI.watch, never photographs