Abstract
Due to computerized diagnosis, risk prediction, and clinical decision support, Artificial Intelligence (AI) has been a central part of modern healthcare. Most AI systems are built upon static models and hand-tuned parameters, which limit efficiency, accuracy, and flexibility. The study suggests OptiCareNet, an optimization-augmented AI model integrating Reinforcement Learning-based Clinical Decision Routing (RL-CDR) for dynamic decision and Particle Swarm Optimization (PSO) for hyperparameter optimization to bridge these limitations. To minimize training loss and enhance diagnostic accuracy, OptiCareNet systematically optimizes neural network architectural parameters such as learning rate, dropout rate, and layer structure. Meanwhile, RL-CDR agents boost real-time therapy planning and response ordering by optimizing treatment trajectories depending on patient state alterations. The MIMIC-IV and UCI Heart Disease datasets were employed to test the framework. Experimental outcomes report improved diagnostic accuracy (98.2%), decreased convergence time (42%), and 35.7% decrease in the rate of erroneous diagnoses compared to conventional deep learning models. The strength of integrating AI with optimization algorithms to provide versatile, efficient, and clinically relevant healthcare solutions is exemplified by OptiCareNet. This approach is scalable and applicable across diagnostic areas such as oncology, critical care, and cardiology.
| Original language | English |
|---|---|
| Title of host publication | Medical Minds and Machine Learning |
| Subtitle of host publication | Transforming Care with AI |
| Publisher | CRC Press |
| Pages | 1-14 |
| Number of pages | 14 |
| ISBN (Electronic) | 9781040983102 |
| ISBN (Print) | 9781041212867 |
| DOIs | |
| Publication status | Published - 1 Jan 2026 |
| Externally published | Yes |
Fingerprint
Dive into the research topics of 'OptiCareNet: An Optimization-Driven AI Framework for Intelligent Clinical Decision Support'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver