Eschrich AI Lab · Translational clinical AI

From molecule
to clinic

AI research across the translational journey from molecular and spatial biology through clinically meaningful prediction to applications in real patient care

Research scope

The Lab follows the full translational arc from biological discovery and multimodal diagnosis to AI in clinical workflows. A central focus is oncology, connecting molecular, spatial, imaging and clinical data around questions that matter for patients.

01 · The translational journey

Covering the translational journey in one pipeline

  1. 01Basic Science
  2. 02Multimodal Diagnostics
  3. 03Clinical Workflows
01Basic Science

Fundamental research into biological signal

Experimental and computational lab research across molecular, tissue and spatial scales to reveal clinically meaningful biology

H&E · multiplex immunofluorescence · spatial features
02Multimodal Diagnostics

Connect modalities for diagnosis

Integrate pathology, imaging, laboratory and clinical data for diagnosis, stratification and clinically useful prediction

PET · CT · biomarkers · multimodal modelling
03Clinical Workflows

Bring AI into clinical workflows

Design and prospectively study physician-supervised AI in real patient care, where patients, clinicians and healthcare systems meet

Voice & LLM systems · safety · usability · clinical utility

02 · Selected projects

AI applied to diverse scientific questions along the pipeline

01

Spatial biology · iCCA

Where is prognostic biology organised?

Investigating the tumour-liver interface through routine H&E, 15-marker multiplex immunofluorescence and thousands of spatial features

Multiplex immunofluorescence of the tumour-liver interface
Multiplex immunofluorescence | Spatial tissue architecture
Whole slide H and E of liver tumour tissue
Whole-slide H&E | Histomorphology
02

Multimodal prediction · NET / PRRT

Can modalities converge on a useful signal?

Combining PET, CT and laboratory biomarkers to study clinically relevant outcome prediction rather than performance metrics in isolation

Three dimensional PET scan with gradient heatmap analysis
3D PET | Gradient-based model interpretation
Multimodal PET CT and biomarker model architecture
PET + CT + biomarkers | Multimodal prediction
03

Clinical AI · ChariCall

Does the system help in practice?

Evaluating a patient-facing voice and LLM workflow for structured pre-visit history taking, with physicians in the loop

Johannes Eschrich in a physician coat at Charité Berlin
Clinical medicine | Charité Berlin
Johannes Eschrich in clinical practice
Implementation in the real clinical workflow

03

Selected publications

04 · Collaborations

Combining medical and technical excellence

Charité Universitätsmedizin BerlinFraunhofer Heinrich Hertz InstituteBerlin Institute of Health at Charité

Interested in collaborating?

I welcome conversations with clinical teams, researchers and builders who want to answer meaningful scientific questions and move translational AI into practice

Get in contact