Evaluation using machine learning delivers first results
The Bio2Treat team of the University Hospital Aachen and Grünenthal GmbH successfully completed the recording of clinical data in November 2022. Thus, clinical data from the clinical examinations and daily measurements from the PainWatch are available for statistical evaluation and analysis using machine learning methods.Initial, simple predictive models have now been developed by the Bio2Treat team. These initial results show that prediction of pain intensity is generally possible. For accurate longer-term prediction, more complex and personalized models are needed. With our work such more complex models can now be applied, which can use long-term correlations from environmental factors to predict pain intensity. Larger amounts of training data are needed to create accurate predictions.Further analyses promise to uncover interesting correlations that can be harnessed for chronic pain therapy.
Schematic representation of machine
Caption: Schematic representation of machine learning. Data from the clinical examination of the test subjects, from the daily records of the PainWatch and from the digital pain diary are used to create a computer model that enables prediction of future pain intensity. Various models from the field of machine learning can be used for this purpose. More complex models from the field of deep learning may also be considered as data collection is completed and larger data sets become available.