Scholarlink

The rapid advancement of artificial intelligence (AI) and machine learning (ML) represents global trends that could significantly enhance the quality of life. In medical terms, the health sector focuses on prevention. In this regard, classification and recurrent artificial neural networks (ANN) with short- and long-term memory, along with other ML algorithms, represent an ideal alternative due to their ability to identify hidden patterns in complex phenomena, such as chronic diseases. Today, the generation of clinical data and its exploitation through intelligent models represents a strategic value for public health; however, the incidence and mortality associated with these diseases continue to increase for vulnerable groups. This study explores the integration of deep models as a predictive diagnostic strategy for common diseases in contaminated areas and indigenous communities of Hidalgo, México. It aims to improve healthcare outcomes through early detection and prevention. By applying the CRISP-DM methodology for data science, among 9 alternative models, the best performing models were as follows: deep ANN for chronic respiratory diseases with f1 score of 99.49% and random forest for diabetes and arterial hypertension with f1 score of 98.7% and 95.47%, respectively. The recursive systolic and diastolic pressure models achieved R2 of 0.9969 and 0.9990, with RMSE of 0.01118 and 0.00555, respectively. This study is considered relevant due to the innovative integration not identified in the reviewed literature.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Predictive Diagnosis of Chronic Diseases Using Deep Learning Approaches for Vulnerable Populations

  • 章节

  • 发表日期：

    2026年

查看全文
  • Silvia Soledad Moreno Gutiérrez,
  • Daniel Tlanepantla Pantoja,
  • Sócrates López Pérez,
  • Héctor Daniel Molina Ruíz

摘要

The rapid advancement of artificial intelligence (AI) and machine learning (ML) represents global trends that could significantly enhance the quality of life. In medical terms, the health sector focuses on prevention. In this regard, classification and recurrent artificial neural networks (ANN) with short- and long-term memory, along with other ML algorithms, represent an ideal alternative due to their ability to identify hidden patterns in complex phenomena, such as chronic diseases. Today, the generation of clinical data and its exploitation through intelligent models represents a strategic value for public health; however, the incidence and mortality associated with these diseases continue to increase for vulnerable groups. This study explores the integration of deep models as a predictive diagnostic strategy for common diseases in contaminated areas and indigenous communities of Hidalgo, México. It aims to improve healthcare outcomes through early detection and prevention. By applying the CRISP-DM methodology for data science, among 9 alternative models, the best performing models were as follows: deep ANN for chronic respiratory diseases with f1 score of 99.49% and random forest for diabetes and arterial hypertension with f1 score of 98.7% and 95.47%, respectively. The recursive systolic and diastolic pressure models achieved R2 of 0.9969 and 0.9990, with RMSE of 0.01118 and 0.00555, respectively. This study is considered relevant due to the innovative integration not identified in the reviewed literature.