Swiss Data Science Center, ETH Zürich and EPFL, Zürich, Switzerland.
The Netherlands Organization for Applied Scientific Research TNO, P.O. Box 2215, 2301 CE, Leiden, The Netherlands.
BMC Med Inform Decis Mak. 2022 Jul 6;22(1):179. doi: 10.1186/s12911-022-01918-2.
Our aim was to develop a machine learning model, using real-world data captured from a connected auto-injector device and from early indicators from the first 3 months of treatment, to predict sub-optimal adherence to recombinant human growth hormone (r-hGH) in patients with growth disorders.
Adherence to r-hGH treatment was assessed in children (aged < 18 years) who started using a connected auto-injector device (easypod™), and transmitted injection data for ≥ 12 months. Adherence in the following 3, 6, or 9 months after treatment start was categorized as optimal (≥ 85%) versus sub-optimal (< 85%). Logistic regression and tree-based models were applied.
Data from 10,929 children showed that a random forest model with mean and standard deviation of adherence over the first 3 months, infrequent transmission of data, not changing certain comfort settings, and starting treatment at an older age was important in predicting the risk of sub-optimal adherence in the following 3, 6, or 9 months. Sensitivities ranged between 0.72 and 0.77, and specificities between 0.80 and 0.81.
To the authors' knowledge, this is the first attempt to integrate a machine learning model into a digital health ecosystem to help healthcare providers to identify patients at risk of sub-optimal adherence to r-hGH in the following 3, 6, or 9 months. This information, together with patient-specific indicators of sub-optimal adherence, can be used to provide support to at-risk patients and their caregivers to achieve optimal adherence and, subsequently, improve clinical outcomes.
我们的目的是开发一种机器学习模型，使用从连接的自动注射器设备和治疗开始后前 3 个月的早期指标中收集到的真实世界数据，来预测生长障碍患者对重组人生长激素（r-hGH）治疗的不依从情况。
对使用连接的自动注射器设备（easypod™）并传输了至少 12 个月注射数据的儿童（年龄 < 18 岁）进行 r-hGH 治疗依从性评估。治疗开始后 3、6 或 9 个月的依从性分为最佳（≥ 85%）和不依从（< 85%）。应用逻辑回归和基于树的模型。
来自 10929 名儿童的数据显示，在最初 3 个月内，基于平均和标准差的依从性、数据传输不频繁、不改变某些舒适设置以及年龄较大时开始治疗的随机森林模型，对预测接下来的 3、6 或 9 个月不依从的风险很重要。敏感性在 0.72 到 0.77 之间，特异性在 0.80 到 0.81 之间。
据作者所知，这是首次尝试将机器学习模型集成到数字健康生态系统中，以帮助医疗保健提供者识别有风险的患者，他们在接下来的 3、6 或 9 个月内可能对 r-hGH 的治疗不依从。这些信息，结合不依从的具体患者指标，可以用于为有风险的患者及其护理人员提供支持，以实现最佳依从性，进而改善临床结局。
