Artificial intelligence and e-SUS APS/PEC: development of predictive models to identify people in need of palliative care in primary health care

Abstract

The early identification of people with palliative care needs remains a challenge for Primary Health Care (PHC), particularly in the context of population ageing, multimorbidity, and increasing complexity of chronic conditions. In this setting, artificial intelligence and predictive models may support the analysis of clinical and healthcare information, helping identify individuals who could benefit from a palliative care assessment. This study aims to analyze the scientific evidence and technical and healthcare possibilities related to the development of artificial intelligence-based predictive models for identifying people with palliative care needs in PHC, considering the potential use of data from the Citizen Electronic Health Record (PEC), part of the Brazilian e-SUS APS strategy. An integrative literature review is proposed, based on studies addressing machine learning, palliative care identification instruments, electronic health records, and digital transformation within the Brazilian Unified Health System (SUS). Evidence concerning predictive algorithms, instruments such as NECPAL and SPICT, clinical data quality, interoperability, and data governance will be considered. The synthesis is intended to support discussion of potential applications, methodological limitations, and requirements for possible implementation in PHC. Artificial intelligence should be regarded as a clinical decision-support resource rather than a substitute for professional assessment. Progress in this field depends on consistent scientific evidence, data quality, model validation, and safeguards for security, equity, and the protection of users' information.