As healthcare systems generate larger volumes of clinical, imaging, and administrative data, a persistent question is how that information can be organized and modeled to support medical decision-making. Records often accumulate faster than systems can standardize or connect them, creating a gap between raw data and usable information. A series of research papers examines this question across several stages of the healthcare data pipeline, from the databases that store medical records to the machine-learning methods built on top of them. In the 2026 paper Research on the Design and Optimization of Automated Data Collection and Visual Dashboard in the Medical Industry, published in the Journal of Computer, Signal, and System Research, automated data collection and visual dashboards are examined as methods for making medical data more accessible in operational settings.
Much of the potential value of healthcare data is held in systems that are large, fragmented, and difficult to query, limiting how readily the information can support downstream analysis. This work treats data infrastructure as a starting point for healthcare AI, based on the premise that analytical models depend on the pipelines that supply their data. In the 2025 paper Utilize the Database Architecture to Enhance the Performance and Efficiency of Large-Scale Medical Data Processing, published in Artificial Intelligence and Digital Technology, database architecture is examined in relation to the performance and efficiency of large-scale medical data processing.
A second strand of the research applies machine learning to cancer-related problems. The 2025 paper Research on the Application of Integrating Medical Data Intelligence and Machine Learning Algorithms in Cancer Diagnosis, published in the International Journal of Engineering Advances, examines how medical-data intelligence and machine-learning algorithms can be combined to support cancer diagnosis. A companion study, Research on Improving the Matching Efficiency between Cancer Patients and Clinical Trials Based on Machine Learning Algorithms, published in the Journal of Medicine and Life Sciences, considers the use of machine-learning methods to improve the process of matching cancer patients with potentially relevant clinical trials.
A third strand addresses the unstructured text found in medical records, where information may appear in free-form notes rather than standardized fields. In Research on Medical Named Entity Recognition Technology Based on Prompt BioMRC Model for Deep NLP Algorithm, published in the International Journal of Big Data Intelligent Technology, a prompt-based model is examined for identifying medical entities in text. A related study, Medical Entity Recognition Based on Bidirectional LSTM-CRF and Natural Language Processing Technology and Its Application in Intelligent Consultation, applies a bidirectional LSTM-CRF approach to medical entity recognition and considers its application in intelligent consultation.
Taken together, the studies can be read as addressing connected stages of healthcare data processing: the infrastructure used to collect and store information, the machine-learning methods applied to cancer diagnosis and clinical-trial matching, and the natural-language methods used to convert medical text into structured data. Each strand focuses on a different part of the process through which medical data may become usable within an analytical system.
The author of this research, Xiangtian Hui, works across software engineering, machine learning, and healthcare data. His research includes medical data infrastructure, cancer informatics, and medical natural-language processing. This combination places his work at the intersection of software engineering and healthcare-related data analysis.
By connecting data engineering, predictive modeling, and medical language processing, this body of work offers one view of how related stages of a healthcare data system can be studied together. The papers point toward a broader research direction in which data infrastructure, machine-learning methods, and language-processing systems are considered as connected components of healthcare AI. Developing and evaluating those connections remains an area of active research.