Department of Global Health, Boston University School of Public Health, 801 Massachusetts Ave, Crosstown Center 3rd Floor, Boston, MA, 02119, USA.
Department of Epidemiological Surveillance, Ministry of Health, Yaoundé, Cameroon.
Sci Rep. 2021 Mar 24;11(1):6713. doi: 10.1038/s41598-021-85987-9.
Although acute respiratory infections are a leading cause of mortality in sub-Saharan Africa, surveillance of diseases such as influenza is mostly neglected. Evaluating the usefulness of influenza-like illness (ILI) surveillance systems and developing approaches for forecasting future trends is important for pandemic preparedness. We applied and compared a range of robust statistical and machine learning models including random forest (RF) regression, support vector machines (SVM) regression, multivariable linear regression and ARIMA models to forecast 2012 to 2018 trends of reported ILI cases in Cameroon, using Google searches for influenza symptoms, treatments, natural or traditional remedies as well as, infectious diseases with a high burden (i.e., AIDS, malaria, tuberculosis). The R and RMSE (Root Mean Squared Error) were statistically similar across most of the methods, however, RF and SVM had the highest average R (0.78 and 0.88, respectively) for predicting ILI per 100,000 persons at the country level. This study demonstrates the need for developing contextualized approaches when using digital data for disease surveillance and the usefulness of search data for monitoring ILI in sub-Saharan African countries.
虽然急性呼吸道感染是撒哈拉以南非洲地区主要的死亡原因，但对流感等疾病的监测大多被忽视。评估流感样疾病（ILI）监测系统的有效性并开发预测未来趋势的方法对于大流行的准备工作非常重要。我们应用并比较了一系列强大的统计和机器学习模型，包括随机森林（RF）回归、支持向量机（SVM）回归、多变量线性回归和 ARIMA 模型，以预测 2012 年至 2018 年喀麦隆报告的 ILI 病例趋势，使用谷歌搜索流感症状、治疗方法、天然或传统疗法以及负担较高的传染病（即艾滋病、疟疾、结核病）。在大多数方法中，R 和 RMSE（均方根误差）在统计学上相似，但 RF 和 SVM 具有最高的平均 R（分别为 0.78 和 0.88），用于预测国家层面每 10 万人的 ILI。本研究表明，在使用数字数据进行疾病监测时需要开发上下文相关的方法，并且搜索数据对于监测撒哈拉以南非洲国家的 ILI 非常有用。
