Development of an AI-Driven Predictive Model for Multiple Myeloma Disease Progression and Treatment Optimization Using Chinese Real-World Data

Multiple myeloma (MM) is a type of blood cancer that affects the bone marrow, where blood cells are made. This disease can be difficult to treat because it varies greatly between patients, and finding the right treatment for each individual is not always easy. In fact, more than 20,000 people in the United States are diagnosed with multiple myeloma each year, and this number is expected to increase. Managing MM involves balancing survival chances, treatment effectiveness, and the potential side effects of treatment.

Currently, doctors use standard statistical models to predict how MM will progress in patients, but these models are not always personalized enough to help doctors make the best treatment decisions for each person. This research aims to develop a new, more accurate way to predict the course of MM using artificial intelligence (AI). Specifically, we will create an AI model that can predict how the disease will progress, how well treatments will work, and what side effects a patient might experience. The AI will be tailored to Chinese patients with MM, using real-world data (RWD) from their health records.

To build this model, we will use advanced AI techniques like deep learning and transformer-based methods, which allow the model to learn from large amounts of data and recognize complex patterns that might be missed by traditional methods. By analyzing sequences of clinical events, changes in health indicators, and the types of treatments patients receive, the AI model will make more personalized predictions. This can help doctors better understand how a patient's disease will progress and what treatments will work best for them.

The research will involve gathering data from real-world health records of Chinese patients with MM, which will be used to train and test the AI model. This process will be done in a way that ensures patient privacy and complies with all ethical standards. Given the higher reliability of data from randomized controlled trials (RCTs), this study intends to use the applied data as the training set for model construction. Subsequently, real-world data from China will be employed as an external validation set to verify the model's generalization ability and select the model with the optimal generalization performance.

In summary, this project aims to make the treatment of multiple myeloma more personalized and effective. By developing a smarter AI tool, doctors will be able to provide more tailored care to MM patients, ultimately improving their chances of survival and quality of life.

多发性骨髓瘤（Multiple Myeloma，MM）是一类累及骨髓的血液系统恶性肿瘤，而骨髓正是人体造血细胞生成的场所。该疾病的患者间异质性极强，为其治疗带来了极大挑战，为每名患者找到适配的治疗方案往往并非易事。事实上，美国每年有超过2万名患者被确诊为多发性骨髓瘤，且这一确诊人数预计还将持续增长。多发性骨髓瘤的临床管理需要在患者生存获益、治疗疗效与潜在治疗不良反应之间寻求平衡。

当前，临床医生多采用标准统计模型预测多发性骨髓瘤的疾病进展，但这类模型往往缺乏足够的个性化程度，难以辅助医生为每名患者制定最优的治疗决策。本研究旨在借助人工智能（Artificial Intelligence，AI）开发一种更为精准的多发性骨髓瘤病程预测方法。具体而言，我们将构建一款AI模型，用以预测疾病进展轨迹、治疗响应效果以及患者可能出现的不良反应。该AI模型将针对中国多发性骨髓瘤患者进行定制化开发，所使用的数据来源于患者的真实世界健康记录（real-world data，RWD）。

为构建该模型，我们将采用深度学习、基于Transformer的方法等先进AI技术，这类技术可使模型从大规模数据中学习并识别传统方法可能遗漏的复杂模式。通过分析临床事件序列、健康指标变化以及患者所接受的治疗类型，该AI模型将生成更具个性化的预测结果，从而帮助医生更精准地预判患者的疾病进展情况，并筛选出最适配的治疗方案。

本研究将从中国多发性骨髓瘤患者的真实世界健康记录中采集数据，用于AI模型的训练与测试。整个数据处理过程将严格保障患者隐私，并符合所有伦理规范。鉴于随机对照试验（randomized controlled trials，RCTs）数据具备更高的可靠性，本研究计划将此类数据用作模型构建的训练集；后续将采用中国地区的真实世界数据作为外部验证集，以验证模型的泛化能力，并筛选出泛化性能最优的模型。

综上，本项目旨在使多发性骨髓瘤的治疗更具个性化与有效性。通过开发一款更智能的AI工具，临床医生将能够为多发性骨髓瘤患者提供更为定制化的诊疗服务，最终提升患者的生存获益与生活质量。