Advancing reliability and specificity of automatic multimodal recognition of pressure and heat pain intensit

点击翻译按钮获取中文摘要

Currently used methods for clinical pain assessment have limited reliability, validity, are time consuming and can only limited applied to patients with restricted communicative verbal abilities. If valid measurement of the pain is not possible, treating the pain may lead to cardiac stress in risk patients, underperfusion of the operating field, over- or under-usage of analgesics and other problems of mistreatment in acute or chronic pain.Main goal of the project proposal is the advancement of pain diagnosis and monitoring of pain states. With the use of multimodal sensor technology and highly effective data classification, reliable and valid automated pain recognition will be possible. To reach this goal the combination of experimental protocols and new powerful methods of data analysis, pattern recognition and machine learning will be a promising strategy for the development of objective pain measurement. Biomedical, visual and audio data will be measured under experimentally controlled conditions in healthy controls. After measurement, the data will be pre-analyzed with a variety of complex filter and decomposition techniques to extract and select meaningful features. These features are the input for a robust automatic recognition of pain intensities in real-time.This is a renewal proposal of the DFG-Project: Advancement and Systematic Validation of an Automated Pain Recognition System on the Basis of Facial Expression and Psychobiological Parameters. It advances the previous project with the following key aspects:1. Generalizability: We will test and improve the generalizability with a complex pain model with extended pain modalities: phasic and tonic, heat and pressure. 2. Response specificity: The specificity of the pain recognition system will be described in comparison to psychosocial stress.3. Assessment modalities: In addition to psychobiological and facial parameters we will assess paralinguistic pain expressions, skin temperature, body movement and other modalities for pain recognition. 4. The trans-temporal reliability of automated pain recognition will be tested by repeating the experimental protocol after (at least) one week. 5. We will compare different person-specific calibration methods, which improve recognition rates.6. All pain classifiers will be adapted for online processing for the advancement of a clinically useful pain monitoring system.