School of Engineering and Applied Sciences, Harvard University, Cambridge, Massachusetts.
Santa Fe Institute, Santa Fe, New Mexico.
Am J Prev Med. 2014 Sep;47(3):341-7. doi: 10.1016/j.amepre.2014.05.020. Epub 2014 Jul 2.
Google Flu Trends (GFT) claimed to generate real-time, valid predictions of population influenza-like illness (ILI) using search queries, heralding acclaim and replication across public health. However, recent studies have questioned the validity of GFT.
To propose an alternative methodology that better realizes the potential of GFT, with collateral value for digital disease detection broadly.
Our alternative method automatically selects specific queries to monitor and autonomously updates the model each week as new information about CDC-reported ILI becomes available, as developed in 2013. Root mean squared errors (RMSEs) and Pearson correlations comparing predicted ILI (proportion of patient visits indicative of ILI) with subsequently observed ILI were used to judge model performance.
During the height of the H1N1 pandemic (August 2 to December 22, 2009) and the 2012-2013 season (September 30, 2012, to April 12, 2013), GFT's predictions had RMSEs of 0.023 and 0.022 (i.e., hypothetically, if GFT predicted 0.061 ILI one week, it is expected to err by 0.023) and correlations of r=0.916 and 0.927. Our alternative method had RMSEs of 0.006 and 0.009, and correlations of r=0.961 and 0.919 for the same periods. Critically, during these important periods, the alternative method yielded more accurate ILI predictions every week, and was typically more accurate during other influenza seasons.
GFT may be inaccurate, but improved methodologic underpinnings can yield accurate predictions. Applying similar methods elsewhere can improve digital disease detection, with broader transparency, improved accuracy, and real-world public health impacts.
谷歌流感趋势（GFT）声称使用搜索查询实时生成具有时效性的、有效的人群流感样疾病（ILI）预测结果，这一技术一经发布便广受赞誉，并在公共卫生领域得到了广泛应用。然而，最近的研究对 GFT 的有效性提出了质疑。
提出一种替代方法，以更好地挖掘 GFT 的潜力，同时为更广泛的数字疾病检测提供辅助。
我们的替代方法于 2013 年开发，该方法可自动选择要监控的特定查询，并在每周新的 CDC 报告的 ILI 数据可用时自动更新模型。使用均方根误差（RMSE）和 Pearson 相关系数来比较预测的 ILI（表示 ILI 就诊患者比例）与随后观察到的 ILI，以判断模型性能。
在 H1N1 大流行（2009 年 8 月 2 日至 12 月 22 日）和 2012-2013 年季节（2012 年 9 月 30 日至 2013 年 4 月 12 日）期间，GFT 的预测值的 RMSE 分别为 0.023 和 0.022（即，如果 GFT 预测一周内 ILI 为 0.061，预计误差为 0.023），相关系数 r 分别为 0.916 和 0.927。而我们的替代方法在同一时期的 RMSE 分别为 0.006 和 0.009，相关系数 r 分别为 0.961 和 0.919。重要的是，在这些重要时期，替代方法每周都能提供更准确的 ILI 预测结果，并且在其他流感季节通常更准确。
GFT 可能不准确，但改进的方法基础可以产生准确的预测。在其他地方应用类似的方法可以提高数字疾病检测的准确性，并提高透明度，产生实际的公共卫生影响。
