other · Frontiers in psychology · la publicación, 14 oct 2026 · gratis
Una universidad china ya recluta estudiantes para un plan de cinco años que tomará datos de la cara y la voz
El programa CAMPUS describe cómo se hará y no presenta resultados. Tampoco dice nada sobre lo que la ley permite en América Latina, Estados Unidos o Canadá.
Versión breve · la versión detallada sigue, unos 7 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
- Estudiantes universitarios de pregrado y posgrado, de 18 a 30 años
- Cuántos
- unos 42,000 estudiantes
- Dónde
- China, siete universidades en siete provincias
- Cuándo
- reclutamiento iniciado en septiembre de 2025; seguimiento de cinco años
- Tipo de estudio
- protocolo de un estudio de cohorte con ensayos planeados; no reporta resultados
- Quién lo hizo
- Centro de Investigación en Salud Mental e Inteligencia Artificial del Segundo Hospital Afiliado de la Universidad Médica de Henan, con siete universidades chinas
- El límite que importa
- Es solo un plan: no hay ningún resultado publicado ni prueba de que funcione
Es un plan, y un plan no puede citarse como prueba de lo que una empresa puede hacer con su rostro o su voz.
En abril de 2025, un comité de ética de la Universidad Médica de Henan, en China, aprobó el primer reclutamiento. Las inscripciones en esa universidad empezaron en septiembre de 2025. Así comienza el registro del programa CAMPUS.
El plan es seguir durante cinco años a unos 42,000 estudiantes de siete universidades chinas, en siete provincias. Cada semestre responderán cuestionarios sobre depresión, ansiedad, insomnio y estrés. Uno se llama PHQ-9 y otro GAD-7.
Un grupo más pequeño dará datos del cuerpo. Hasta 400 estudiantes por centro cada año podrían someterse a electroencefalograma, resonancia magnética, medición del ritmo cardíaco y seguimiento de los ojos. Las grabaciones de voz y de expresión facial son para todo el grupo, cada semestre. La sangre, la orina, las heces y la saliva aparecen como muestras posibles, pero todavía no se recogen.
El programa también describe ocho ensayos separados con intervenciones como terapia cognitivo-conductual, música, ejercicio, realidad virtual y biofeedback. Ninguno ha comenzado: el artículo dice que los ocho siguen en etapa de planificación.
Sobre la inteligencia artificial, el texto es claro. Los modelos se construirán y probarán fuera de línea. No decidirán un diagnóstico, ni la respuesta ante el riesgo de suicidio, ni quién entra a un ensayo, ni qué tratamiento recibe cada estudiante. Cualquier respuesta positiva en el cuestionario de ideación suicida activa una alerta que revisa una persona capacitada.
El artículo no informa ningún resultado: ni cuántos estudiantes mejoraron, ni si los modelos funcionan. Tampoco menciona leyes, reguladores ni tribunales de ningún país. Es un plan, y un plan no puede citarse como prueba de lo que una empresa puede hacer con su rostro o su voz.
Los estudiantes son de siete universidades chinas, elegidas por colaboraciones de investigación y no al azar. No representan a todos los estudiantes de China, y mucho menos a alguien en México, Colombia o Texas. Los permisos y consentimientos que describe rigen allí.
El artículo sí explica cómo se cuidarán las grabaciones y los datos. Se guardarán con un código en lugar del nombre, en sistemas con acceso limitado, y solo pasarán revisiones antes de compartirse. Sobre eso, y sobre lo que usted podría exigir donde vive, este documento no responde.
Qué significa para usted
Lo que se describe es un plan: no hay resultados, ni efecto probado, ni modelo que haya pasado una prueba. Para usted, en México, Colombia o Texas, ese documento no dice qué puede exigir a una empresa que use su rostro o su voz. Si le preocupa que alguien grabe a otra persona, la pregunta que sirve es qué ley local cubre esa grabación, y quién responde ante ella.
Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887
Quién pagó: El trabajo fue financiado por fuentes gubernamentales chinas y provinciales, principalmente la Fundación Nacional de Ciencias Naturales de China y programas de la provincia de Henan, y el artículo afirma que los financiadores no tuvieron papel alguno en el diseño del estudio, la recolección de datos, el análisis, la interpretación ni la preparación del manuscrito; no se mencionan proveedores de equipos o software.
No tome esto como consejo médico profesional.
Versión detalladaLos pasajes copiados del artículo, las ilustraciones y cada fuente con cuánto leímos de ella · 1398 palabras · unos 7 minLeerla →Cerrar
Un programa chino quiere ponerle sensores a la salud mental de 42,000 universitarios. Todavía no empieza.
El protocolo describe cómo se reunirían cuestionarios, voz, rostro, imágenes del cerebro y muestras biológicas para entrenar un modelo. No reporta un solo resultado.
Imaginemos a un estudiante de 20 años frente a una pantalla. Contesta un cuestionario sobre su ánimo, su sueño y su estrés. En otra sala le colocan electrodos en la cabeza, le toman una imagen del cerebro, le graban la voz mientras lee un texto en voz alta y le filman la cara mientras responde preguntas abiertas. Eso es lo que el programa CAMPUS planea pedir a miles de estudiantes en siete universidades de China.
El artículo es un protocolo, es decir, el plano de un estudio que todavía no ha dado resultados. Lo firma un equipo del Centro de Investigación en Salud Mental e Inteligencia Artificial del Segundo Hospital Afiliado de la Universidad Médica de Henan, con investigadores de siete universidades chinas, y lo pagaron la Fundación Nacional de Ciencias Naturales de China y programas de la provincia de Henan. Los autores declararon no tener conflictos comerciales.
Piense en la fila de una clínica universitaria: pocos psicólogos, muchos alumnos. Los autores escriben que la depresión, la ansiedad, el insomnio y la ideación suicida entre universitarios crecen más rápido de lo que pueden atender los servicios de apoyo de los campus, limitados en recursos1. El plan quiere cubrir ese hueco con medición a gran escala.
La cifra es grande: unos 42,000 estudiantes al inicio, unas 6,000 personas por universidad, seguidas una vez por semestre durante cinco años, nueve rondas de seguimiento después de la primera. Con una pérdida estimada del 15% anual, quedarían unos 18,600 al final. El artículo aclara que esas cifras son metas de planeación, no resultados.
Lo que el sistema mediría es amplio. Junta cuestionarios de depresión, ansiedad, insomnio, estrés y riesgo suicida con mediciones del cuerpo y del cerebro: electroencefalograma, resonancia magnética, espectroscopia funcional por infrarrojo cercano, variabilidad del ritmo cardiaco, seguimiento ocular, análisis de la expresión facial, potenciales evocados, acústica de la voz2. A eso se suma una parte de muestras biológicas para estudios de múltiples capas moleculares, que aún no está activada2.
La tecnología que se quiere construir tiene nombre técnico y se puede explicar sin él. Un modelo de inteligencia artificial, del tipo llamado Transformer, tomaría todos esos datos a la vez y buscaría patrones que indiquen riesgo, según lo define el propio cuestionario, no un diagnóstico clínico3. Encima se usaría una técnica llamada SHAP para señalar qué dato pesó en cada predicción3. El modelo se desarrollaría y evaluaría fuera de línea, y su salida no decidiría un diagnóstico, ni la disposición ante un riesgo suicida, ni la elegibilidad para un ensayo, ni la asignación al azar4.
Ocho ensayos aleatorizados están planeados dentro del mismo programa: terapia cognitivo-conductual, terapia interpersonal y de ritmos sociales, ejercicio físico, música personalizada, biorretroalimentación, realidad virtual, arte expresivo y una intervención de regulación del estrés por cinco sentidos. Cada uno con su propio criterio de entrada y su propio grupo de comparación. Los ocho siguen en etapa de planeación: no han comenzado reclutamiento, ni asignación al azar, ni entrega de intervenciones5.
El diseño tiene una cautela que vale la pena mirar de cerca. Cualquier respuesta positiva en la escala de ideación suicida dispara una alerta electrónica que revisa personal capacitado, y la evaluación del riesgo la hace un profesional de salud mental, no el modelo6. El programa distingue tres niveles de respuesta, desde contacto en 24 horas hasta traslado a urgencias. La inteligencia artificial queda fuera de esa cadena6.
Hay otro límite explícito. Ser detectado por el cuestionario o por el modelo no mete a nadie a un ensayo: hace falta una evaluación específica, un consentimiento aparte y una asignación al azar hecha por un sistema independiente de la plataforma7. La participación es voluntaria y se puede retirar el consentimiento en cualquier momento8.
Sobre los datos personales, el protocolo dice tres cosas concretas. Las grabaciones de voz, los videos del rostro, las imágenes cerebrales y los datos genómicos no se depositarán en repositorios públicos9. Cada componente sensible exige un consentimiento separado10. Y negarse a uno de ellos no afecta la participación en el resto del estudio11. Lo que el artículo no dice es cómo se guardan, se comparten o se destruyen esas grabaciones en la práctica.
Para que algo de esto ocurra faltan pasos. Ninguna muestra biológica se recolectará hasta que los procedimientos, los materiales de consentimiento y las enmiendas éticas estén aprobados12. Cada universidad nueva necesita el visto bueno de su comité de ética local antes de activarse13. El reclutamiento comenzó en la Universidad Médica de Henan en septiembre de 2025, y el resto de las sedes está por arrancar.
Los propios autores enumeran lo que este diseño no puede sostener. Las universidades se eligieron por colaboraciones de investigación, no por sorteo de una lista nacional, así que la muestra no representa a China. La muestra es mayoritariamente de etnia han. La validación es interna: entre participantes, en el tiempo y dejando fuera una sede a la vez, y los autores escriben que no la llamarán validación externa independiente14. Excluyeron a estudiantes con diagnóstico previo de depresión mayor o ansiedad generalizada. Las mediciones fisiológicas se hacen con aparatos distintos en cada centro. Y las atribuciones del modelo no se interpretarán como prueba de un mecanismo biológico ni de causalidad15.
Así lo lee weeklyAI: un protocolo es una promesa escrita, y las promesas se juzgan por lo que dicen de sí mismas. Este documento es inusualmente franco sobre sus límites, lo cual es una buena señal y no una prueba de nada. Lo que un lector de México, Bogotá o Los Ángeles debe retener es más simple: aquí no hay ningún hallazgo que citar. Nadie ha demostrado todavía que juntar voz, rostro y cerebro mejore la detección de la depresión, ni que estas ocho intervenciones funcionen.
Esa distancia entre el tamaño del plan y la ausencia de resultados es la noticia. Un programa que aspira a 42,000 estudiantes y ocho ensayos puede terminar publicando que el modelo multimodal no superó a un cuestionario de nueve preguntas. También puede publicar lo contrario. Hasta entonces, cualquier titular que atribuya a CAMPUS una mejora en la salud mental de los universitarios chinos se estaría adelantando a los hechos. Esta lectura se equivocaría si los ensayos publican resultados con grupos de comparación claros y el modelo muestra ventaja medible sobre los cuestionarios solos.
Otra cosa que weeklyAI espera ver, y que aquí no se puede verificar: cuando alguien entrega su cara y su voz a un estudio, la pregunta práctica no es qué promete el consentimiento, sino quién contesta si algo sale mal, en cuánto tiempo y ante qué instancia ajena al propio equipo. El protocolo describe consentimientos separados y la opción de negarse. No describe plazos de respuesta ni canales de queja. Ese es el hueco donde suelen perderse los derechos en papel.
Un detalle que conviene vigilar desde lejos: los estudiantes que rechacen la resonancia o la muestra de sangre quedarán en una posición distinta a la de quienes acepten todo. El protocolo dice que eso no afecta su participación11. Lo que no puede decir es cómo se sentirá ese rechazo dentro de un campus donde el estudio es visible. Si los documentos publicados informan cuántos rechazan y por qué, y tratan ese rechazo como un dato normal, la preocupación se disuelve. Si no lo informan, queda la duda.
Para el lector de este lado del mundo, la utilidad es indirecta y real. Este programa no rige en su país ni le da derechos nuevos. Pero marca hacia dónde camina la conversación: universidades, hospitales y gobiernos que quieren detectar el riesgo emocional con aparatos en lugar de con personas. Cuando en su ciudad alguien proponga algo parecido, las preguntas que este protocolo deja abiertas sirven tal cual: qué se mide, quién decide con esa medición, qué pasa si usted dice que no, y quién responde por sus grabaciones.
La pantalla del principio sigue encendida. El estudiante contesta, la voz se graba, la imagen se guarda. Lo que este documento permite saber es que, por ahora, nadie ha publicado qué encontraron esos datos. Esa es la pregunta que puede llevar a la siguiente reunión donde le pidan su cara, su voz o su sangre: ¿y los resultados, dónde están?
De dónde sale cada dato de contexto, y cuánto leímos de cada documento
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - el artículo del que trata esta nota — el artículo completo — el pasaje: "University students face a rising prevalence of depression, anxiety, insomnia, and suicidal ideation that exceeds the capacity of conventional, resource-constrained campus support services."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Multidimensional screening integrates psychological self-report (PHQ-9, GAD-7, ISI, BSI-CV, and others), physiological measurements (EEG, MRI, fNIRS, HRV, eye tracking, facial-expression analysis, ERP, voice acoustics), and an exploratory biospecimen component for multi-omics profiling"
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - el artículo del que trata esta nota — el artículo completo — el pasaje: "A Transformer-based multimodal AI framework with SHAP-based feature attribution will be developed and internally evaluated for questionnaire-defined risk stratification and exploratory treatment-response prediction."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - el artículo del que trata esta nota — el artículo completo — el pasaje: "All models will be developed and evaluated offline. Their outputs will not determine clinical diagnosis, suicide-risk disposition, trial eligibility, randomized allocation, treatment selection, intervention intensity, or protocol modification."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - el artículo del que trata esta nota — el artículo completo — el pasaje: "All eight embedded intervention trials remain in the planning stage and have not commenced participant recruitment, enrolment, randomization, or intervention delivery."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - el artículo del que trata esta nota — el artículo completo — el pasaje: "AI models will be developed and evaluated offline; their outputs will not be used for real-time safety management, clinical risk assignment, or disposition and will not replace or delay professional assessment or emergency action."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Cohort screening or AI-generated model outputs will not automatically enroll a participant in an intervention or determine trial selection or randomized allocation."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Participation is voluntary, and participants may withdraw consent for continued participation at any time."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Raw voice recordings, facial-expression videos, imaging data, genomic data, and other data carrying a material risk of re-identification will not be deposited in public repositories."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Voice recording, facial-expression video recording, physiological and neuro-functional assessments, MRI, and future biospecimen collection for genetic and other multi-omics analyses will be subject to separate component-specific consent or consent options"
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Participants may decline or discontinue any optional component without affecting their continued participation in the core cohort."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - el artículo del que trata esta nota — el artículo completo — el pasaje: "No biospecimen collection will begin until the relevant standard operating procedures, consent materials, and required ethics amendments have been approved."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Each additional participating university will obtain approval from its responsible ethics committee before site activation, participant enrolment, or data collection."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - el artículo del que trata esta nota — el artículo completo — el pasaje: "These procedures constitute internal validation and will not be described as independent external validation."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - el artículo del que trata esta nota — el artículo completo — el pasaje: "SHAP analyses of held-out predictions will be used to describe feature contributions but will not be interpreted as evidence of biomarker validity, biological mechanism, or causality."
Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887
Quién pagó: El trabajo fue financiado por fuentes gubernamentales chinas y provinciales, principalmente la Fundación Nacional de Ciencias Naturales de China y programas de la provincia de Henan, y el artículo afirma que los financiadores no tuvieron papel alguno en el diseño del estudio, la recolección de datos, el análisis, la interpretación ni la preparación del manuscrito; no se mencionan proveedores de equipos o software.
No tome esto como consejo médico profesional.
other · Frontiers in psychology · the paper, 14 Oct 2026 · free
China Plans a Five-Year Mental-Health Study of 42,000 Students. No Results Exist Yet.
A protocol describes questionnaires, brain scans and eight therapy trials still in planning. It says nothing about what any law allows with a student's face or voice.
Short version · the longer version follows, about 7 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
- University students in China
- How many
- about 42,000 students at seven Chinese universities
- Where
- China
- When
- enrolment began September 2025; five-year study
- Kind of study
- survey
- Who did it
- Henan Medical University and partner universities
- The limit that matters
- No results yet; the eight treatment trials have not begun.
The protections here are promises, not rights a student could enforce.
In April 2025, an ethics committee at Henan Medical University approved the first recruitment and mental-health assessment of students at that university. Enrolment there began in September 2025. What followed is a plan, not a result.
The plan is a five-year study of about 42,000 students at seven Chinese universities, from Henan in the centre to Hainan in the south. Each student would answer questionnaires once a semester, nine times after the first. The main one is the PHQ-9, nine questions that score depressive symptoms from 0 to 27. Others ask about anxiety, sleep, stress and suicidal thoughts.
A smaller group would give more. Up to 400 students per university per year would sit for brain-wave and brain-imaging recordings, heart-rate measurement, eye tracking, voice recordings and video of their faces. Blood, urine, stool and saliva are listed as possible samples. None has been collected yet.
Eight experiments are described, each testing one approach against a comparison group: talk therapy, a sleep-and-routine therapy, music, exercise, virtual reality, art, biofeedback and a five-sense stress program. Each would recruit 420 students. The article says all eight remain in planning. None has begun.
The artificial intelligence is described as a set of computer models built and tested away from students, on data already gathered. The article states these models will not decide a diagnosis, a response to suicide risk, who joins a trial or which treatment anyone receives. Any positive answer on the suicide questionnaire triggers a human safety review, whatever a model says.
No results are reported. The article gives no effect, no accuracy figure and no count of students helped or harmed. It does say that keeping or destroying data and samples will follow the law where the study runs.
Two things stay open. The article does say that data and samples get coded study numbers, that the key linking those numbers to a person is kept apart and locked away, and that only authorized staff may see them. It adds that raw recordings, images and genetic data will not be placed in public repositories, and that keeping or destroying them will follow the approved protocol, the consent forms, university policy and the law. The students are at seven Chinese universities chosen through research partnerships, so the plan says nothing about rights or procedures in Latin America, the United States or Canada.
What this means for you
For now, there is nothing here about your rights, and no result to act on. What the plan does say is that the key linking a student's coded number to their name is kept apart and locked away, and that raw face, voice and genetic records stay out of public archives. Ask your own authorities where a student's recordings may go.
Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887
Who paid: The work was funded by Chinese government and provincial sources, chiefly the National Natural Science Foundation of China and Henan Province programs, and the article states the funders had no role in study design, data collection, analysis, interpretation or manuscript preparation; no equipment or software lenders are mentioned.
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 · 1381 words · about 7 minRead it →Close
China Plans a Five-Year Study of 42,000 Students, With AI Reading Faces, Voices and Blood
The CAMPUS protocol describes a screening system that combines questionnaires, brain scans and genetic data. No results exist yet: the eight treatment trials have not begun.
Consider what a student at one of seven Chinese universities would be asked to do. Once a semester for five years, she fills in a set of standard questionnaires about mood, worry, sleep and stress. If she is one of those invited for the fuller battery, she also sits for a brain scan, a heart-rhythm recording, an eye-tracking session, a voice recording while she reads a passage aloud, and a video of her face as she answers open questions. She may be asked for blood, urine, stool and saliva.
That is the shape of CAMPUS, described in a study protocol in Frontiers in Psychology; the authors date the current version of the master protocol 5 September 2026. The program is a five-year cohort across seven Chinese universities, with an operational target of about 6,000 students at each, roughly 42,000 at the start. It also plans eight separate randomized trials of psychological and lifestyle interventions, none of which has begun. The article reports no findings: it states plainly that it describes the protocol and does not report participant-level or aggregate results from the cohort.
The problem the authors name is capacity. University students face a rising prevalence of depression, anxiety, insomnia and suicidal ideation that exceeds what conventional, resource-constrained campus support services can handle1. Their answer is a funnel: screen many students with questionnaires, add measurements a questionnaire cannot give, and let software sort the results. The software is a Transformer-based multimodal framework, a type of AI model built to weigh several kinds of input at once, with SHAP-based feature attribution, a method for showing which inputs pushed a given prediction2. The inputs are the questionnaires, the physiological measurements, and the exploratory biospecimen component for multi-omics profiling3.
The safety architecture is where the protocol is most explicit, and it is worth reading closely. All models will be developed and evaluated offline, and their outputs will not determine clinical diagnosis, suicide-risk disposition, trial eligibility, randomized allocation, treatment selection, intervention intensity, or protocol modification4. The AI will not be used for real-time safety management and will not replace or delay professional assessment or emergency action5. Screening results will not automatically enroll anyone in a trial or decide who is randomized where6. A positive answer on the suicide-ideation questionnaire triggers an immediate electronic alert to trained site staff and the designated site safety officer, and each alert gets prompt human review, with three response levels written into the protocol.
On consent, the design is layered. Voice recording, facial-expression video, physiological and neuro-functional assessments, MRI, and future biospecimen collection for genetic and other multi-omics analyses are each subject to separate component-specific consent7. A participant may decline or stop any optional component without affecting her place in the core cohort8, and participation is voluntary throughout, with consent withdrawable at any time9. Raw voice recordings, facial videos, imaging and genomic data will not be deposited in public repositories10.
Much remains unfinished, and the protocol says so. All eight embedded trials remain in the planning stage and have not begun recruitment, enrolment, randomization or delivery11. No biospecimen collection will begin until the relevant procedures, consent materials and ethics amendments are approved12. Each additional university needs its own ethics committee approval before it activates13. The authors also list the limits: the seven universities were chosen through research collaborations rather than national sampling, so the cohort is not nationally representative; the sample is described as predominantly Han Chinese; MRI and biospecimen participation may attract a systematically different subset of students; devices differ by centre; and excluding students with diagnosed psychiatric disorders limits the examination of the full continuum of symptom severity.
Other work in this field suggests why a protocol, rather than a result, is the honest unit here. According to the summary of a review of AI mental-health tools for adolescents, the only part of it available because the full paper is behind a subscription, the searches found 11,167 records and included 103 sources, of which 95 were primary studies14. Predictive or diagnostic performance was reported in 83 of those studies, but symptom-level effectiveness in only one15. No primary study provided a dedicated prospective AI-safety evaluation16, and no study evaluated sustained routine-care scale-up or implementation cost as a principal outcome17. The review's own conclusion was that the field is dominated by prediction and internal validation, while symptom-level benefit, prospective safety and scale-up remain largely untested18.
What students say about such tools is also on record. According to the summary of a study of Nigerian adults who used AI mental-health chatbots for at least four weeks, the only part available, participants valued round-the-clock availability and a non-judgmental space but struggled with the lack of genuine human reciprocity19. Prolonged use generated attachment anxiety about the app being discontinued or data being breached, compounded by the absence of human escalation pathways20. The authors of that study concluded that ethical deployment requires culturally adaptive design, transparent algorithmic boundaries and mandatory human handoff protocols21.
Here is weeklyAI's reading. The pattern to watch is what happens to intimate records once they are gathered at scale. A program that films faces, records voices, maps brains and sequences genes through a single digital platform will hold exactly the kind of material that others find valuable, and the pressure to share, reuse or transfer it does not disappear because the consent form is careful. This reading would be wrong if the program published, in plain language and over the full five years, who can open each type of record, confirmed that no commercial partner ever receives any of it, and showed through independent checks that practice matched the documents. Until then, the useful question for anyone invited into a study like this is a written one: who holds the face, voice and biological records, and can a company ever receive them?
A second pattern is the authority a score carries. A number that flags a student as at risk looks measured, yet the cutoffs, the questionnaires and the training data are human choices, and a student who lands near a line may be treated differently because of where the line fell. This reading would be wrong if the program published how its thresholds were chosen, showed that flagged students fare no worse than unflagged ones, and demonstrated that the scores hold when tested at universities that had no part in building them. If a school, clinic or employer ever shows you an AI risk score about your mental health, ask what data went in, who set the cutoff, and whether a person reviewed the result before anyone acted on it.
The protections here are promises, not rights a student could enforce. Ethics approvals, consent forms and internal safety procedures govern the program; the article describes no law a Chinese student could invoke, and nothing in it says what a student in Mexico, Brazil or Ohio could demand of anyone. This reading would be wrong if students in the program turn out to have a specific legal route to challenge misuse of their data or a mishandled alert, and an outside body with power to penalize has actually used it.
The protocol itself points to what must happen before any of this becomes evidence. Each trial needs its own ethics approval, a dated intervention manual, a statistical analysis plan and a public registration before the first participant is randomized. The validation the authors plan is internal only, by design: they state that these procedures constitute internal validation and will not be described as independent external validation22. Even the explanations the AI gives about its own predictions are fenced off: SHAP analyses will describe feature contributions but will not be interpreted as evidence of biomarker validity, biological mechanism or causality23.
So the thing to hold on to is the questionnaire, the video, the blood draw. None of them has yet told anyone whether a student was helped. When the first results appear, the question to bring is the plain one: did the students who were screened and sorted end up better off than those who were not, and who checked?
Where each piece of context comes from, and how much of it we read
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - the article this story is about — the whole article — the passage: "University students face a rising prevalence of depression, anxiety, insomnia, and suicidal ideation that exceeds the capacity of conventional, resource-constrained campus support services."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - the article this story is about — the whole article — the passage: "A Transformer-based multimodal AI framework with SHAP-based feature attribution will be developed and internally evaluated for questionnaire-defined risk stratification and exploratory treatment-response prediction."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - the article this story is about — the whole article — the passage: "Multidimensional screening integrates psychological self-report (PHQ-9, GAD-7, ISI, BSI-CV, and others), physiological measurements (EEG, MRI, fNIRS, HRV, eye tracking, facial-expression analysis, ERP, voice acoustics), and an exploratory biospecimen component for multi-omics profiling"
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - the article this story is about — the whole article — the passage: "All models will be developed and evaluated offline. Their outputs will not determine clinical diagnosis, suicide-risk disposition, trial eligibility, randomized allocation, treatment selection, intervention intensity, or protocol modification."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - the article this story is about — the whole article — the passage: "AI models will be developed and evaluated offline; their outputs will not be used for real-time safety management, clinical risk assignment, or disposition and will not replace or delay professional assessment or emergency action."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - the article this story is about — the whole article — the passage: "Cohort screening or AI-generated model outputs will not automatically enroll a participant in an intervention or determine trial selection or randomized allocation."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - the article this story is about — the whole article — the passage: "Voice recording, facial-expression video recording, physiological and neuro-functional assessments, MRI, and future biospecimen collection for genetic and other multi-omics analyses will be subject to separate component-specific consent or consent options"
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - the article this story is about — the whole article — the passage: "Participants may decline or discontinue any optional component without affecting their continued participation in the core cohort."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - the article this story is about — the whole article — the passage: "Participation is voluntary, and participants may withdraw consent for continued participation at any time."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - the article this story is about — the whole article — the passage: "Raw voice recordings, facial-expression videos, imaging data, genomic data, and other data carrying a material risk of re-identification will not be deposited in public repositories."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - the article this story is about — the whole article — the passage: "All eight embedded intervention trials remain in the planning stage and have not commenced participant recruitment, enrolment, randomization, or intervention delivery."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - the article this story is about — the whole article — the passage: "No biospecimen collection will begin until the relevant standard operating procedures, consent materials, and required ethics amendments have been approved."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - the article this story is about — the whole article — the passage: "Each additional participating university will obtain approval from its responsible ethics committee before site activation, participant enrolment, or data collection."
- Dhamayanti M, Susaldi S, Hanifa F, Agustini Y, Emilia O, Hardinsyah H, et al. (2026). Artificial intelligence-enabled mental health applications for adolescents: a scoping review of safety, effectiveness and clinical implementation. BMJ Open. 10.1136/bmjopen-2026-125586 — only the abstract - the full paper is behind a subscription — the passage: "The database searches identified 11 167 records, and 1 additional evidence synthesis was identified and assessed during revision; 103 sources were included, comprising 95 primary studies and 8 evidence syntheses."
- Dhamayanti M, Susaldi S, Hanifa F, Agustini Y, Emilia O, Hardinsyah H, et al. (2026). Artificial intelligence-enabled mental health applications for adolescents: a scoping review of safety, effectiveness and clinical implementation. BMJ Open. 10.1136/bmjopen-2026-125586 — only the abstract - the full paper is behind a subscription — the passage: "Predictive or diagnostic performance was reported in 83 studies (87.4%), but symptom-level effectiveness in only 1 (1.1%)."
- Dhamayanti M, Susaldi S, Hanifa F, Agustini Y, Emilia O, Hardinsyah H, et al. (2026). Artificial intelligence-enabled mental health applications for adolescents: a scoping review of safety, effectiveness and clinical implementation. BMJ Open. 10.1136/bmjopen-2026-125586 — only the abstract - the full paper is behind a subscription — the passage: "No primary study provided a dedicated prospective AI-safety evaluation."
- Dhamayanti M, Susaldi S, Hanifa F, Agustini Y, Emilia O, Hardinsyah H, et al. (2026). Artificial intelligence-enabled mental health applications for adolescents: a scoping review of safety, effectiveness and clinical implementation. BMJ Open. 10.1136/bmjopen-2026-125586 — only the abstract - the full paper is behind a subscription — the passage: "No study evaluated sustained routine-care scale-up or implementation cost as a principal outcome."
- Dhamayanti M, Susaldi S, Hanifa F, Agustini Y, Emilia O, Hardinsyah H, et al. (2026). Artificial intelligence-enabled mental health applications for adolescents: a scoping review of safety, effectiveness and clinical implementation. BMJ Open. 10.1136/bmjopen-2026-125586 — only the abstract - the full paper is behind a subscription — the passage: "The field is dominated by prediction and internal validation, whereas symptom-level benefit, prospective safety, workflow integration, scale-up and sustainability remain largely untested."
- Isaiah US, Adili-George CF, Naibo AT, George I, Useh EM, Effiong JE. (2026). Lived experiences of relationships to generative AI mental health chatbots among adults. Discover Psychology. https://doi.org/10.1007/s44202-026-01001-1 — only the abstract - the full text could not be fetched — the passage: "Relational Ambivalence ; participants valued 24/7 availability and non-judgmental space yet struggled with perceived lack of genuine human reciprocity"
- Isaiah US, Adili-George CF, Naibo AT, George I, Useh EM, Effiong JE. (2026). Lived experiences of relationships to generative AI mental health chatbots among adults. Discover Psychology. https://doi.org/10.1007/s44202-026-01001-1 — only the abstract - the full text could not be fetched — the passage: "Fragile Dependency ; prolonged use generated attachment anxiety about potential app discontinuation or data breaches, compounded by absence of human escalation pathways."
- Isaiah US, Adili-George CF, Naibo AT, George I, Useh EM, Effiong JE. (2026). Lived experiences of relationships to generative AI mental health chatbots among adults. Discover Psychology. https://doi.org/10.1007/s44202-026-01001-1 — only the abstract - the full text could not be fetched — the passage: "Findings suggest ethical deployment in LMICs requires culturally adaptive design, transparent algorithmic boundaries, and mandatory human handoff protocols to prevent dependency in unsupported contexts."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - the article this story is about — the whole article — the passage: "These procedures constitute internal validation and will not be described as independent external validation."
- Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887 - the article this story is about — the whole article — the passage: "SHAP analyses of held-out predictions will be used to describe feature contributions but will not be interpreted as evidence of biomarker validity, biological mechanism, or causality."
Yang, G., Liu, R., Wang, C. et al. (2026). Comprehensive artificial intelligence-assisted mental health program for university students in China (CAMPUS): study protocol for a multi-center prospective cohort with planned embedded randomized intervention trials. Frontiers in Psychology. https://doi.org/10.3389/fpsyg.2026.1877887
Who paid: The work was funded by Chinese government and provincial sources, chiefly the National Natural Science Foundation of China and Henan Province programs, and the article states the funders had no role in study design, data collection, analysis, interpretation or manuscript preparation; no equipment or software lenders are mentioned.
Do not take this as professional medical advice.

