Biotechnology — Predictive computational models in personalized medicine research — Part 2: Requirements and recommendations for implementing computational models in clinical integrated decision support systems
Introduction: The combination of patient-derived data, such as genomics and lifestyle information, with patient registries and databases holds immense potential for improving medical care. One such potential lies in computer- based personalized models capable of providing precise, patient specific, predictions. Predictive and prognostic models hold great potential to support clinical decision making in personalized medicine and could ultimately facilitate a paradigm shift to a more personalized form of treatment.[1] Various computational models (see 3.10) have been devised to assist with the decision-making process. Prognostic models are mathematical tools that relate a person's current characteristics and individual risk factors to the likelihood of a particular future outcome or disease course. Examples include estimating the probability of disease progression in an already ill patient or predicting whether an individual is likely to develop a disease in the future based on its own risk profile (risk stratification). Prognostic models can incorporate one or many current characteristics (multivariable). When validated, these models provide clinicians with objective estimates of the probability of treatment response, possible complications or quality-of-life-evolution that complements other clinical information. Predictive modelling, on the other hand, uses available data to forecast likely future responses and effects of a specific treatment or intervention. This can be used to identify p
Scope: This document specifies requirements and recommendations for designing, developing, and establishing integrated clinical decision support systems (CDSS) for research purposes in personalized medicine. This document can be used as an implementation guideline for setting up computational modelling workflows to enable such systems. It addresses data quality, formatting and handling, as well as the processes generating such data, and the set-up, validation, simulation, storing and sharing of computational models used for this specific purpose in personalized medicine. This includes recommendations and rules for configuration, descriptions, annotations, interoperability, integration, access and provenance of such data and derived models in an interpretable and evidence-based manner. This document also specifies how to integrate these rules with clinical trials execution applying standard operating procedures. Furthermore, requirements and recommendations for data used to construct or required for validating such models are addressed. This document does not apply for computational models used for diagnostic or therapeutic purposes in standard clinical practice, outside of a formal research or investigational setting.