1. Multimodal and multi-omics AI for precision diagnosis, treatment, and whole-course management of major diseases
This area investigates how medical images, clinical information, physiological measurements, behavioural signals, and molecular or multi-omics data can be integrated to support precision diagnosis, prognostic assessment, and personalised whole-course management of major diseases.
Primary scientific and translational topics in this research area are:
multimodal screening, diagnosis, and precise disease phenotyping using pathology, retinal imaging, CT, MRI, ultrasound, facial or behavioural video, clinical records, and physiological measurements;
AI biomarkers for disease risk, progression, prognosis, complications, recurrence, and treatment response;
integration of imaging phenotypes with genomics, transcriptomics, single-cell data, epitranscriptomics, and other molecular information;
foundation models, medical vision-language models, and data-efficient learning for clinical scenarios with limited labelled data;
generalisation across hospitals, medical imaging machinespopulations, and disease prevalence, including domain adaptation, uncertainty estimation, causal learning, and unsupervised learning; and
clinically interpretable evidence that complements, rather than reproduces, conventional diagnostic and risk-assessment systems.
Existing member foundations in this area
Members contribute strengths in prospective and multicentre AI for childhood vision screening, computational pathology for precision oncology, quantitative cardiovascular imaging for functional and risk assessment, intelligent prenatal ultrasound analysis for congenital heart disease, and single-cell epitranscriptomic modelling for uncovering disease trajectories and regulatory mechanisms. These focuses connect population screening and precise diagnosis with prognosis and biological discovery.
Collaboration opportunities
Opportunities for collaboration include longitudinal cohorts, multicentre external validation, treatment-response studies, pathology-radiology integration, image-omics linkage, rare disease datasets, foundation-model adaptation, and prospective evaluation of AI-supported clinical workflows.
2. Data-driven mechanism discovery and prioritisation of intervention targets for major diseases
This area uses AI to investigate disease mechanisms across molecular, cellular, tissue, and organ system scales. AIHB will integrate single-cell and spatial omics, epitranscriptomics, digital pathology, medical imaging, brain atlases, genetics, and drug-perturbation data to identify reproducible cell states, regulatory pathways, tissue microenvironments, and network-level alterations involved in disease development and progression.
The goal is to transform complex biomedical data into testable mechanistic hypotheses, candidate biomarkers, potential therapeutic targets, and treatment-response signatures. Graph learning, complex-network modelling, and multimodal data integration will connect genes, cells, tissue architecture, brain systems, drugs, and clinical phenotypes, enabling discoveries to be examined across biological scales and independently validated.
Primary scientific and translational topics for this area are:
multiscale disease-mechanism discovery using single-cell, spatial, and epitranscriptomic data, digital pathology, medical imaging, and brain atlases;
identification of disease-associated cell states, regulatory pathways, and tissue microenvironments, with an initial focus on tumour immunity and treatment resistance, and on brain and neurodegenerative diseases;
graph and network modelling to connect genes, cells, tissue architecture, brain systems, drugs, and clinical phenotypes;
cross-dataset validation of mechanistic signals to distinguish reproducible biological findings from dataset-specific associations;
AI-guided prioritisation of candidate biomarkers, therapeutic targets, drug sensitivities, and treatment-response signatures; and
independent validation of computational findings in collaboration with pathology, spatial-omics, wet-lab, and clinical teams to generate testable mechanistic and intervention hypotheses.
Existing member foundations in this area
Members have established research in single-cell epitranscriptomics and rare-cell analysis, computational pathology, biological and brain-network modelling, and drug-target prediction. This cross-scale expertise supports the study of how molecular and cellular changes interact with tissue microenvironments and neural systems. It provides a foundation for mechanism-led research in tumour immunity and treatment resistance, and in brain and neurodegenerative diseases.
Collaboration opportunities
AIHB welcomes collaboration with multi-omics and neuroscience laboratories, biobanks, pathology and clinical teams, pharmaceutical R&D groups and biomedical-technology partners. Joint projects will connect computational discovery with independent datasets, tissue analysis, spatial profiling, laboratory experiments, and clinical validation, turning AI-generated findings into testable mechanistic and therapeutic hypotheses.
3. Digital phenotyping, treatment monitoring, and rehabilitation management
This area investigates how speech, language, facial behaviour, movement, physiological measurements, and longitudinal interaction data can be used to characterise changes in patients’ behavioural, psychological, and functional states during treatment, follow-up, and rehabilitation. Its core objective is to develop intelligent technologies that support longitudinal monitoring, functional assessment, and personalised management while remaining safe, interpretable, and suitable for real-world healthcare and rehabilitation settings.
Primary scientific and translational topics in this area are:
construction of multimodal digital phenotypes reflecting psychological, behavioural, and functional states using speech, language, facial behaviour, movement, physiological measurements, and interaction history;
modelling longitudinal changes in patient states during treatment, follow-up, and rehabilitation, rather than only single-time-point state recognition or generic emotion classification;
multimodal monitoring and assistive assessment of treatment response, symptom changes, pain, fatigue, adherence, and functional recovery;
large language model-assisted interviews, digital humans, conversational agents, and embodied interactive systems for standardised information collection, supportive interaction, and follow-up monitoring of patient states;
AI-assisted rehabilitation assessment and management for stroke, upper-limb impairment, fractures, cerebral palsy, and other physical-function limitations;
gamified and virtual-reality rehabilitation, wearable or environmental sensing systems, smart splints, and other assistive devices.
Existing member foundations in this area
AIHB has established an integrated foundation in affective computing, multimodal state assessment, clinical speech and interview research, human-computer interaction, intelligent rehabilitation, and digital health technologies. These capabilities are supported by the SIP Affective Computing and Interactive Health platform and the Taicang industry joint laboratory. Together, these resources support research on longitudinal changes in patients’ behavioural, psychological, and functional states and real-world evaluation in treatment monitoring, rehabilitation assessment, and personalised management.
Collaboration opportunities
The Centre welcomes collaboration with psychiatry and psychology departments, rehabilitation hospitals, community health organisations, clinical follow-up and health-management teams, special education institutions, medical and rehabilitation-device companies, digital health service providers, and organisations interested in responsible, human-centred AI.
1. Multimodal and multi-omics AI for precision diagnosis, treatment, and whole-course management of major diseases
This area investigates how medical images, clinical information, physiological measurements, behavioural signals, and molecular or multi-omics data can be integrated to support precision diagnosis, prognostic assessment, and personalised whole-course management of major diseases.
Primary scientific and translational topics in this research area are:
Existing member foundations in this area
Members contribute strengths in prospective and multicentre AI for childhood vision screening, computational pathology for precision oncology, quantitative cardiovascular imaging for functional and risk assessment, intelligent prenatal ultrasound analysis for congenital heart disease, and single-cell epitranscriptomic modelling for uncovering disease trajectories and regulatory mechanisms. These focuses connect population screening and precise diagnosis with prognosis and biological discovery.
Collaboration opportunities
Opportunities for collaboration include longitudinal cohorts, multicentre external validation, treatment-response studies, pathology-radiology integration, image-omics linkage, rare disease datasets, foundation-model adaptation, and prospective evaluation of AI-supported clinical workflows.
2. Data-driven mechanism discovery and prioritisation of intervention targets for major diseases
This area uses AI to investigate disease mechanisms across molecular, cellular, tissue, and organ system scales. AIHB will integrate single-cell and spatial omics, epitranscriptomics, digital pathology, medical imaging, brain atlases, genetics, and drug-perturbation data to identify reproducible cell states, regulatory pathways, tissue microenvironments, and network-level alterations involved in disease development and progression.
The goal is to transform complex biomedical data into testable mechanistic hypotheses, candidate biomarkers, potential therapeutic targets, and treatment-response signatures. Graph learning, complex-network modelling, and multimodal data integration will connect genes, cells, tissue architecture, brain systems, drugs, and clinical phenotypes, enabling discoveries to be examined across biological scales and independently validated.
Primary scientific and translational topics for this area are:
Existing member foundations in this area
Members have established research in single-cell epitranscriptomics and rare-cell analysis, computational pathology, biological and brain-network modelling, and drug-target prediction. This cross-scale expertise supports the study of how molecular and cellular changes interact with tissue microenvironments and neural systems. It provides a foundation for mechanism-led research in tumour immunity and treatment resistance, and in brain and neurodegenerative diseases.
Collaboration opportunities
AIHB welcomes collaboration with multi-omics and neuroscience laboratories, biobanks, pathology and clinical teams, pharmaceutical R&D groups and biomedical-technology partners. Joint projects will connect computational discovery with independent datasets, tissue analysis, spatial profiling, laboratory experiments, and clinical validation, turning AI-generated findings into testable mechanistic and therapeutic hypotheses.
3. Digital phenotyping, treatment monitoring, and rehabilitation management
This area investigates how speech, language, facial behaviour, movement, physiological measurements, and longitudinal interaction data can be used to characterise changes in patients’ behavioural, psychological, and functional states during treatment, follow-up, and rehabilitation. Its core objective is to develop intelligent technologies that support longitudinal monitoring, functional assessment, and personalised management while remaining safe, interpretable, and suitable for real-world healthcare and rehabilitation settings.
Primary scientific and translational topics in this area are:
Existing member foundations in this area
AIHB has established an integrated foundation in affective computing, multimodal state assessment, clinical speech and interview research, human-computer interaction, intelligent rehabilitation, and digital health technologies. These capabilities are supported by the SIP Affective Computing and Interactive Health platform and the Taicang industry joint laboratory. Together, these resources support research on longitudinal changes in patients’ behavioural, psychological, and functional states and real-world evaluation in treatment monitoring, rehabilitation assessment, and personalised management.
Collaboration opportunities
The Centre welcomes collaboration with psychiatry and psychology departments, rehabilitation hospitals, community health organisations, clinical follow-up and health-management teams, special education institutions, medical and rehabilitation-device companies, digital health service providers, and organisations interested in responsible, human-centred AI.