From Gynecologic Oncology Clinical Practice to AI-assisted model
Prof. David O’Malley
Country : USA
Official Title : Professor
Department :
Institute : The Ohio State Uni-versity
Speaker CVNavigating Beyond Boundaries: Advancement Toward Durable Outcomes in Endometrial Cancer
Navigating the evolving landscape of endometrial cancer requires strategic treatment sequencing to achieve durable disease control. Prof. O’malley’s esteemed lecture will examine the latest risk stratification criteria, highlighting their prognostic relevance and role in guiding individualized therapy decisions across disease stages. A structured review of the current treatment continuum, including surgical intervention, line of systemic therapies, and immuno-oncology approaches, which underscore the importance of sequencing therapies to optimize outcomes. Emphasis will be placed on real-world experience with PD-1 inhibitors, illustrating clinical considerations that influence therapy prioritization, including biomarker status, patient characteristics, and timing within the treatment pathway. Through case discussions, participants will get to explore practical decision-making processes, demonstrating how strategic integration of immunotherapy at key junctures contributes to sustained disease control. By combining evidence-based insights with experiential learning, this session aims to equip clinicians with a comprehensive framework for navigating complex treatment pathways, enhancing long-term outcomes, and advancing personalized care in endometrial cancer.
張正昌 Prof. Cheng-Chang Chang
Country : Taiwan
Official Title : Professor
Department :
Institute : 中華民國婦癌醫學會
Speaker CVMy path to be a full professor
This presentation by Dr. Cheng-Chang Chang shares a clinical researcher’s journey in gynecologic oncology, illustrating how clinical challenges inspired translational research. The initial motivation stemmed from limitations in cervical cancer screening, especially for glandular lesions (adenocarcinomas), which are less detectable via traditional Pap smears. A nationwide multicenter study led by the Taiwanese Gynecologic Oncology Group demonstrated that methylation biomarkers (e.g., SOX1, POU4F3, ZNF582, PAX1, PTPRR) could effectively triage atypical glandular cells (AGC) and improve detection of CIN3+ lesions, thus reducing overtreatment. Notably, biomarker consistency between physician-collected and self-collected samples supports the potential for broader screening access via self-sampling. Further, the research expanded to explore DNA methylation as a diagnostic tool in endometrial and ovarian cancers. Using bioinformatics and immunohistochemistry, the studies identified molecular subtypes, inflammatory pathways, and possible links between endometriosis and ovarian carcinogenesis (EAOCs). This work exemplifies a cycle of identifying clinical problems, leveraging research collaborations and data platforms, and translating findings into novel screening strategies. The presentation advocates for clinician-led research, highlighting the value of integrating clinical insight with molecular techniques to address unmet needs in gynecologic cancer screening and diagnosis.
李宛珊 Dr. Wan-Shan Li
Country : Taiwan
Official Title : Professor
Department :
Institute : 奇美醫院
Speaker CVIn depth interpretation and application of NGS
Next-generation sequencing (NGS) has become an indispensable tool in gynecologic oncology, uncovering key genetic alterations such as BRCA1/2 mutations, homologous recombination deficiency (HRD), and microsatellite instability (MSI). These biomarkers guide treatment strategies including PARP inhibitors and immunotherapy, and have reshaped precision oncology. Yet, clinical implementation of NGS is challenged by sample quality, testing variability, interpretive complexity, and reimbursement limitations. Artificial intelligence (AI) provides new opportunities to address these challenges. By integrating genomic data with clinical records, imaging, and pathology, AI-assisted models can predict HRD or BRCA status, stratify prognosis, and anticipate therapeutic response. This presentation will review the current applications and limitations of NGS in gynecologic oncology, followed by emerging AI-based approaches that complement molecular testing. Together, these advances highlight how the convergence of NGS and AI may accelerate the transition toward truly personalized cancer care.
張啟昌 Prof. Chi-Chang Chang
Country : Taiwan
Official Title : Professor
Department :
Institute : 中山醫學大學
Speaker CVAI assisted medical decision making
Artificial intelligence (AI) has rapidly become a transformative force in medicine, offering new opportunities for data-driven, evidence-based decision making. In this lecture, Professor Chi-Chang Chang will discuss the role of AI-assisted medical decision making within the broader context of multidisciplinary research. Over the past two decades, his work has focused on integrating diverse fields—including clinical medicine, data science, and healthcare policy—to address complex medical challenges. AI technologies, when combined with interdisciplinary expertise, have the potential to improve diagnostic accuracy, optimize treatment planning, and enhance patient outcomes. However, their implementation also raises important questions regarding ethics, trust, and contextual appropriateness. This presentation will explore both the opportunities and limitations of AI in supporting medical decision making, highlighting the necessity of collaboration across domains to ensure that AI-driven solutions are not only technically feasible but also clinically meaningful and socially responsible.
郭至恩 Prof. Chih En Kuo
Country : Taiwan
Official Title : Professor
Department :
Institute : 國立中興大學
Speaker CV從婦癌臨床實務邁向AI輔助模型
卵巢囊腫的良惡性鑑別診斷是婦產科臨床常見難題，因腫瘤種類繁多且特徵相似，增加了超音波診斷的不確定性。我們提出一套基於深度學習的快速準確輔助分辨卵巢囊腫的良惡性的方法。我們採用十個知名卷積神經網路模型（如AlexNet、GoogleNet、ResNet）進行遷移學習訓練。為確保結果穩定性，重複隨機抽取訓練和驗證資料十次，以平均值作為最終評估。訓練完成後，選取準確率與分類時間比最高的三個模型進行集成學習，並應用Grad-CAM技術視覺化決策結果。研究結果顯示，單一CNN模型的最高平均準確率、靈敏度和特異性分別為90.51±4.36%、89.77±4.16%和92.00±5.95%。集成分類器的相應指標為92.15±2.84%、91.37±3.60%和92.92±4.00%，在所有評估指標上均優於單一分類器，且標準差更小，顯示更高的穩定性和穩健性。綜合考量資料量、多樣性和驗證策略，本方法優於先前研究。未來將導入應用於臨床環境。
top