Clinical researcher · Heidelberg
I ask clinical questions, annoy datasets until they answer, and occasionally recruit AI agents to help.
Clinical Medicine × Psychosomatic Medicine × Mental Health × Pain × Scientific AI
I like new knowledge, new places, new challenges, and—apparently—new notebooks.
Clinically trained across medicine, psychiatry and psychosomatic medicine, I work at the intersection of human health, complex data and scientific AI.
I am particularly interested in turning strong clinical datasets into testable questions, interpretable models and reusable research systems.
Currently in Heidelberg. Frequently found somewhere between a clinical question and a Python terminal.
Research snapshot →
THE OPERATING SYSTEM
I tend to work on several questions in parallel. Switching between problems often helps me see connections I would miss if I stayed inside one project for too long.
Multitasking? Maybe. Parallel processing sounds more academic.
I enjoy research most when a problem requires me to learn something I did not already know.
The topic changes. The habit does not: I like finding relationships, structures and logic between things that initially look unrelated.
ONE QUESTION, MANY WINDOWS
Why can similar biological or psychological experiences become very different symptoms, behaviours and healthcare needs in different people?
RESEARCH WORLD
I study chronic pain as a heterogeneous, multidimensional human experience shaped by adversity, distress, perception, coping and functional burden.
Pain: multidimensional. Dataset: also multidimensional. My RAM: less multidimensional.
RESEARCH WORLD
Adolescent mental health unfolds across young people, parents, families, schools and pathways to help.
RESEARCH WORLD
Is the health that medicine measures always the same as the health that people experience? Interviews, scales and clinical-entry data each reveal only part of the picture.
absence of medicalization ≠ absence of suffering
RESEARCH WORLD
Can scientific work itself be redesigned? That question eventually led me to agents, reusable workflows and the Research OS.
AND HARD DRIVE
What do difficult life experiences have to do with chronic pain?
MANUSCRIPT / DOCTORAL RESEARCHopen notes +Can adversity-related, psychological and pain-related information reveal a shared latent burden dimension?
MANUSCRIPTopen notes +How do pain-related psychological characteristics and emotional symptoms interact within adolescent–parent dyads?
MANUSCRIPTopen notes +When a symptom exists, but medicine barely sees it.
ACTIVE RESEARCHopen notes +RAG supports evidence retrieval; it does not validate the findings.
Can scientific research workflows become modular, auditable and reusable without outsourcing scientific judgement?
SYSTEM BUILDING / ACTIVEopen notes +How can multiple AI agents make scientific decisions without becoming an untraceable black box?
SYSTEM BUILDING / ACTIVEopen notes +The interesting part is not adding more agents. It is deciding what each agent is allowed to know, do and decide.
RESEARCH MODES
SELECTED WORK
Four papers chosen for what they demonstrate—not for the length of the bibliography.
BMC Medicine · 2026
What it askedCan we move from ‘these factors are associated with school aversion’ toward a more explicit causal model of how they may relate?
Humanities and Social Sciences Communications · 2026
What it askedHow do family characteristics and parenting stress sit inside a connected system of children’s emotional and behavioural difficulties?
Pain Research and Management · 2026 · ACCEPTED
What it askedCan adolescent pain perception and coping be represented as something richer than a single pain score?
BMC Public Health · 2025 · 25:2994
What it askedCan interpretable machine learning reveal clinically meaningful patterns associated with adolescent self-harm?
METHODS I HAVE ACTUALLY USED IN PUBLISHED WORK
Tools are more convincing when they leave evidence behind.
Causal modeling of school aversion in psychiatrically referred adolescents: a DoWhy-based analysisBMC Medicine
DOI ↗Network analysis of the associations among parental age, primary caregivers, parenting stress, and children’s emotional–behavioral difficultiesHumanities and Social Sciences Communications
DOI ↗Development and validation of a multidimensional psychometric scale for assessing pain perception and coping strategies among adolescentsPain Research and Management · Accepted
Decoding the adolescent non-suicidal self-injury: understanding with interpretable machine learning insightsBMC Public Health · 25:2994
DOI ↗SKETCHBOOK PAGES
Research has occasionally taken me places. European railways have occasionally taken me somewhere else.
I love travelling. European railways have occasionally interpreted this as a collaborative decision-making process. Several trips ended somewhere I did not intend to visit. I generally visited it anyway.
A VERY SHORT LAB CAREER
During my first cell-biology practical, I somehow managed to kill the entire class’s cells.
I subsequently developed a healthy respect for people who can keep cells alive.
PUZZLE CORNER
Here is a puzzle instead.
1/3Three sealed notebooks are labelled Pain, AI and Both. Every label is wrong. You may inspect one page. Which notebook do you open?
Best ways to start a conversation: an interesting dataset · a strange clinical question · a puzzle I should not spend the next hour solving
SCIENTIFIC AI, WITH SUPERVISION
I dislike repeating tasks that a machine could reasonably learn to do.
Not because repetitive work is beneath me. Because every hour spent renaming files, checking the same formatting rule, rebuilding the same literature table or retracing an old analysis is an hour not spent asking a better question.
How much of scientific work can be made reusable, auditable and modular—without outsourcing scientific judgment?
RESEARCH OS
Raw data → audit → hypotheses → literature → analysis → comparison → figures → manuscript → review → provenance. The path loops whenever reality disagrees.
HUMAN JUDGMENT stays in the loop.
PROOF, NOT BUZZWORDS
CORE CAPABILITIES
Each track is connected to work or an output—not a proficiency percentage.
From a vague clinical problem to an operational hypothesis, data model, analysis plan and interpretable endpoint.
Model linked people, latent structure, connected systems and explicit causal assumptions.
Build and interrogate models so their signals remain clinically discussable.
Interpretation > leaderboard score, where appropriate.I am interested in AI as research infrastructure, not only as a prediction model.
Framework names describe exposure; the evidence is in the workflow architecture.EVIDENCE GRAPH
PRINT / SCREEN
The serious version, still on one notebook spread.
SEM · APIM · Psychometrics · Network Analysis · RGCCA · Simulation
DoWhy · DAGs · observational causal inference
scikit-learn · SHAP · interpretable ML · predictive modeling
LangGraph · AutoGen · RAG · Vector DB · LoRA · PEFT · Multi-Agent
R · Python · SPSS · AMOS
DO NOT OPEN
What I ultimately want to build
Highly developed medical systems contain specialist expertise, clinical workflows, accumulated experience, decision structures, training systems, data and technology. These cannot simply be copied into remote or resource-limited settings.
Can we distill advanced medical knowledge, workflows and decision support into forms that can travel?
My long-term dream is to help make good medicine less dependent on postcode.
Especially in psychiatry.01How can large clinical cohorts become discovery engines rather than paper factories?
02Can AI agents help scientists generate and falsify hypotheses systematically?
03How do psychosocial experiences become measurable biological or clinical burden?
04Why do medical systems see some symptoms clearly and barely see others?
05Can we connect pain, mental health, neuroinflammation and brain health using multimodal longitudinal data?
06What happens when medicine, causal inference and Scientific AI are designed together from the beginning?
07
Probably enough projects.
COLLABORATION
I am particularly interested in collaborations where the clinical question comes first and the method is allowed to evolve.
I am also interested in teams with strong clinical data that are beginning to build their AI / computational research capability.
ESPECIALLY INTERESTED IN
Agent: ‘I found one more.’Mengmeng: ‘Email me.’