AI-Powered Early Disease Forecasting: A New Frontier in Individualized Care

An editorial published in the journal Intelligent Medicine outlines a series of artificial intelligence (AI) approaches designed to analyze dynamic changes in health data for early disease forecasting and individualized care. The authors identify both the potential and current limitations of these methods.

Publication Details

• Title: "Dynamics-driven medical big data mining: dynamic approaches to early disease forecasting and individualized care"
• Journal: Intelligent Medicine, Volume 6, Issue 1
• Publication Date: February 2026
• Access: Open Access
• Authors:
  • Lu Wang (Tianjin Medical University)
  • Han Lyu (Beijing Friendship Hospital, Capital Medical University)
  • Bin Sheng (Shanghai Jiao Tong University)

Core Concept and Technical Approaches

The editorial discusses the potential for AI to detect subtle, early changes in health data before clinical symptoms fully manifest. Several technical frameworks are described:

This method aims to detect impending disease transitions by monitoring increases in fluctuations and correlations within biomolecular networks. The editorial notes prior applications include flagging gene-expression instability in influenza infection before symptoms appear and identifying genomic tipping points in tumor progression, with reported prediction accuracies exceeding 80%.

This approach transforms molecular data into edge networks to assess critical transitions using a single patient's longitudinal data, without requiring a control group. In transcriptomic applications, this method has reportedly achieved area-under-the-curve (AUC) values greater than 0.9.

The editorial presents evidence that combining mechanistic physiological knowledge with deep learning can improve clinical utility. An example cited is physiology-informed long short-term memory (LSTM) networks for type 1 diabetes management, which reportedly reduced the mean absolute error in blood-glucose prediction to 35.0 mg/dL, compared to 79.7 mg/dL for traditional simulators.

The editorial also describes:

• Temporal graph neural networks applied to electronic health records (EHRs), reporting a 10–15% improvement in diagnosis prediction accuracy on the MIMIC-III dataset.
• Dynamic graph models derived from functional MRI for predicting treatment outcomes in tinnitus.
• Transformer-based architectures trained on longitudinal EHRs for forecasting multi-disease risks.

Author Statement

"These dynamics-driven approaches are designed to augment, not replace, clinical expertise. They provide timely early-warning signals that empower proactive intervention, moving medicine from reactive treatment to genuine prevention, while preserving the irreplaceable role of human judgment in complex medical decision-making."

— Bin Sheng, Corresponding Author and Professor at Shanghai Jiao Tong University

Identified Limitations and Challenges

The editorial identifies several current challenges for the field:

• Technical Hurdles: Data heterogeneity and missing values can produce false positives in critical transition detection. Current methods are noted to excel at identifying statistical associations but cannot reliably distinguish correlation from causation without incorporating medical domain knowledge and experimental validation.
• Interpretability: Full transparency in deep learning architectures is described as not yet achieved, with tools like SHAP and LIME providing only partial explanations.
• Ethical and Regulatory Concerns: These include privacy risks in federated learning architectures and algorithmic bias when models trained on specific populations are deployed in underrepresented groups.

Priorities for Future Work

The editorial identifies two main priorities for advancing the field:

1. Multimodal Integration: Fusing diverse data types—including omics, medical imaging, electronic health records, and wearable device data—through advanced AI methods like Transformers, graph neural networks, and causal inference.
2. Rigorous Prospective Validation: Conducting well-designed prospective clinical trials and real-world deployment studies across diverse populations and healthcare settings.

The editorial is described as serving as a reference and practical roadmap for clinicians, researchers, and healthcare leaders working at the intersection of medicine and artificial intelligence.