Originally published in CTIMES

The research team led by Distinguished Professor Cheng-Yu Chen at Taipei Medical University has successfully developed Taiwan’s first Clinical Decision Support System for Shared Decision-Making (CDSS-SDM) for lung cancer. This project, part of the Big Data Precision Medicine AI System for Lung Cancer, aims to achieve early prevention, accurate diagnosis, and precise treatment. The system facilitates shared decision-making between doctors and patients and has been honored with the latest FUTEX Future Tech Award and the National Innovation Award.

Lung cancer is the leading cause of cancer-related deaths in Taiwan, and only early diagnosis and precise treatment can overcome it. Lung cancer treatment decisions require multiple considerations—early clinical diagnosis relies on high-precision imaging, while treatment and drug selection must take into account various factors such as oncogenic gene mutations.

From the very first day of imaging diagnosis, patients are in a race against time. In the critical first 10 days, at least four test results must be reviewed and discussed with experienced physicians to determine the best treatment approach. Leveraging big data and artificial intelligence can facilitate early decision-making, enhance medical efficiency, and help achieve the vision of precision clinical treatment for lung cancer.

The “Lung Cancer Clinical Intelligent Decision Support System” was developed through a collaboration between Taipei Medical University and the Ministry of Science and Technology, bringing together interdisciplinary biomedical experts to enhance AI-powered big data applications. By leveraging an innovative AI lung cancer model, the system assists in interpreting clinical CT scans and digital pathology images, integrating clinical data and genetic information. This breakthrough platform enables early lung cancer diagnosis and facilitates a personalized precision treatment model based on shared decision-making between doctors and patients.

The team has developed a “Fully Automated Low-Dose CT Lung Cancer Gene Mutation Prediction Model”, which can automatically detect tumors from over 300 CT images, precisely segment and analyze them, determine tumor types and potential gene mutations, and generate automated reports with lung nodule management recommendations.

Furthermore, the team has integrated CT-based predictions with clinical big data, using lung adenocarcinoma patient datasets and automated machine learning methods to establish predictive models for brain metastasis, prognosis, and drug response. When a new case is detected in CT imaging, the system can immediately predict brain metastasis risk and provide drug selection recommendations.

Taipei Medical University, in collaboration with Taiwan’s leading medical technology company AetherAI Technology, has developed a cutting-edge technique that enables high-speed searching and interpretation of cancer cells within vast whole-slide digital pathology images.

Building on this innovation, the team has further developed a “Fully Automated Digital Lung Adenocarcinoma Pathology Gene Mutation Prediction and Drug Selection Model.” This model can rapidly and automatically annotate cancerous regions and predict mutations in the epidermal growth factor receptor (EGFR)—one of the most common gene mutations in lung adenocarcinoma.

By integrating pathology data with clinical datasets from thousands of patients, this breakthrough facilitates early and precise drug selection. The research team is currently pursuing domestic and international patents for these groundbreaking advancements.

Furthermore, the team has leveraged breakthrough technologies to develop the “Pathology Report Natural Language Processing (NLP) Automatic Interpretation and Drug Recommendation System” and the “Comprehensive Genomic Drug Recommendation Model for Lung Adenocarcinoma.”

By utilizing AI-driven natural language processing (NLP), the system can automatically generate drug recommendations—both insurance-covered and self-paid options—that are associated with higher survival rates, simply by analyzing a patient’s pathology report. Additionally, the model links treatment outcomes with survival data, helping to identify the most effective treatment options for patients with similar conditions and the best prognoses.

For patients with advanced lung cancer who are ineligible for surgery or facing metastasis and recurrence, the system automatically matches their condition with global clinical trial databases, providing them with potential access to cutting-edge experimental treatments and new therapeutic opportunities.