AI-based Analysis of multimodal medical data

Intelligently connecting sensor data for different modalities (e.g. blood pressure, heart rate, metabolic parameters such as glucose, and medical history data) can help healthcare personnel to better understand and interpret the psychophysiological interrelationships in the human body. For example, connecting a person’s heart rate to movement parameters makes it possible to obtain a significantly better overall picture of their fitness and physical condition.

Fusing electrophysiological measurement data with camera images for emotion analysis provides more reliable information about someone’s overall psychophysiological condition. In addition, sensor data fusion can achieve better signal quality, while employing several inexpensive sensors can help reduce costs.

AI-based longitudinal prognoses and predictive modeling

With the help of near-sensor, AI-based predictive models, it is possible to measure and evaluate longitudinal patient data (e.g. medical history data, lab results, biosignals) over longer time periods, and with only minimal strain on the patient. Healthcare personnel can thus spot correlations with disease progressions more quickly and predict the condition’s potential residual effects.

Physicians can visualize the models in real time and look at these together with the patient to point out, for example, the effects that medication or lifestyle changes have on the course of therapy.

Your advantages with us as a partner
To process and evaluate the data volumes for specific target groups, for many years we have applied both conventional analysis methods and AI-based approaches, such as the training of deep neural networks (DNNs).

Our work here builds on an interdisciplinary network of other research institutions, universities and hospitals, access to global databases and a high-performance deep-learning computing cluster. We attach great importance to compliance with the European General Data Protection Regulation (GDPR).

What you can expect from us

• Requirements analysis in direct dialogue with you
• Literature review
• Recording of data sets in studies of test subjects
• Evaluation and analysis of multimodal data sets
• Development of customer-specific AI algorithms and licensing of existing algorithms for data analysis (e.g. R-wave detection and recognition of arrhythmia using textile-integrated electrodes for ECGs)
• Benchmarking using independent test data and medical reference devices
• Final discussion