Using Markov Decision Processes to Optimize Breast Biopsy Decision Making

使用马尔可夫决策过程来优化乳腺活检决策

• 批准号: 7385606
• 负责人: Oguzhan Alagoz
• 金额: $ 18.53万
• 依托单位: UNIVERSITY OF WISCONSIN-MADISON
• 依托单位国家: 美国
• 财政年份: 2008
• 资助国家: 美国
• 起止时间: 2008-02-01 至 2010-01-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/7385606
• 关键词: Affect; Age; Anxiety; Artificial Intelligence; Benign; Biopsy; Breast; Breast Cancer Detection; Breast Diseases; Cancer Patient; Computer Simulation; Death Rate; Decision Analysis; Decision Making; Diagnosis; Disease; Early Diagnosis; Future; Healthcare Systems; Invasive; Malignant Neoplasms; Mammography; Modeling; Pain; Patients; Population; Probability; Procedures; Process; Public Health; Reporting; Research; Risk; Risk Factors; Screening procedure; Skin Cancer; Societies; Techniques; Uncertainty; Ursidae Family; Woman; age group; base; computer based statistical methods; cost; improved; malignant breast neoplasm; programs

DESCRIPTION (provided by applicant): Project Summary Early diagnosis through screening mammography is the most effective means of decreasing the death rate from breast cancer. While mammography is inexpensive, the interventional procedures that result from detected abnormalities (both false and true positives) increase the cost of this population-based screening program significantly. In fact, breast biopsy actually delivers a benign result in over 80% of cases making it the most costly per capita component of a breast cancer screening program. If a mammogram reports a suspicious finding, then a biopsy is required to decide whether an abnormality is in fact a breast cancer. A false positive mammogram exposes the patient to the anxiety, pain, and possible complications while the health care system bears the cost of potentially unnecessary biopsies. Our previous research has developed a probabilistic computer model called the Mammography Bayesian Network (MBN) that calculates the risk of breast disease based on demographic risk factors and mammography findings. The objective of this research is to optimize the biopsy decisions for breast-cancer patients such that the early diagnosis of invasive breast cancer is improved while unnecessary invasive procedures are minimized. We will calibrate our previously developed MBN, to accurately calculate the risk of breast cancer based on demographic risk factors and mammography findings. We will use Markov decision processes, an advanced decision analysis technique that is used for decision- making under uncertainty, to find the optimal probability thresholds for the decision to perform breast biopsy for patients with different age groups. We will determine whether these optimal probability thresholds change with patient age. Relevance of this research to Public Health: The proposed research will improve the interpretation of screening mammography, the most effective means of decreasing the death rate from breast cancer, which affects millions of women in the US. Any improvement in screening mammography will reduce the costs of unnecessary biopsies to the society.

通过筛查性乳房X光检查进行早期诊断是降低乳腺癌死亡率的最有效手段。虽然乳房X线摄影是廉价的，但由于检测到的异常（假阳性和真阳性）而导致的介入手术显著增加了这种基于人群的筛查计划的成本。事实上，乳腺活检实际上在超过80%的病例中提供了良性结果，使其成为乳腺癌筛查计划中人均成本最高的组成部分。如果乳房X光检查报告可疑发现，则需要进行活检以确定异常是否实际上是乳腺癌。假阳性乳房X光检查使患者暴露于焦虑、疼痛和可能的并发症，而医疗保健系统则承担可能不必要的活检费用。我们以前的研究已经开发了一种称为乳腺X线摄影贝叶斯网络（MBN）的概率计算机模型，该模型根据人口统计学风险因素和乳腺X线摄影结果计算乳腺疾病的风险。本研究的目的是优化乳腺癌患者的活检决策，以提高浸润性乳腺癌的早期诊断，同时最大限度地减少不必要的侵入性手术。我们将校准我们以前开发的MBN，以根据人口统计学风险因素和乳房X光检查结果准确计算乳腺癌的风险。我们将使用马尔可夫决策过程（一种用于不确定性决策的先进决策分析技术）来寻找不同年龄段患者进行乳腺活检决策的最佳概率阈值。我们将确定这些最佳概率阈值是否随患者年龄而变化。 这项研究与公共卫生的相关性：拟议的研究将改善对乳房X光检查筛查的解释，这是降低乳腺癌死亡率的最有效手段，乳腺癌影响着美国数百万女性。乳腺X光筛查的任何改进都将减少不必要的活检对社会的成本。