Interim evaluation of Google AI forecasting for COVID-19 compared with statistical forecasting by human intelligence in the first week

This article has been Reviewed by the following groups

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Abstract

Background

Since June, Google (Alphabet Inc.) has provided forecasting for COVID-19 outbreak by artificial intelligence (AI) in the USA. In Japan, they provided similar services from November, 2020.

Object

We compared Google AI forecasting with a statistical model by human intelligence.

Method

We regressed the number of patients whose onset date was day t on the number of patients whose past onset date was 14 days prior, with information about traditional surveillance data for common pediatric infectious diseases including influenza, and prescription surveillance 7 days prior. We predicted the number of onset patients for 7 days, prospectively. Finally, we compared the result with Googles AI-produced forecast. We used the discrepancy rate to evaluate the precision of prediction: the sum of absolute differences between data and prediction divided by the aggregate of data.

Results

We found Google prediction significantly negative correlated with the actual observed data, but our model slightly correlated but not significant. Moreover, discrepancy rate of Google prediction was 27.7% for the first week. The discrepancy rate of our model was only 3.47%.

Discussion and Conclusion

Results show Googles prediction has negatively correlated and greater difference with the data than our results. Nevertheless, it is noteworthy that this result is tentative: the epidemic curve showing newly onset patients was not fixed.

Article activity feed

  1. SciScore for 10.1101/2020.12.16.20248358: (What is this?)

    Please note, not all rigor criteria are appropriate for all manuscripts.

    Table 1: Rigor

    NIH rigor criteria are not applicable to paper type.

    Table 2: Resources

    Software and Algorithms
    SentencesResources
    By contrast, Google provided their prediction for 4 weeks from November 19, 2020.
    Google
    suggested: (Google, RRID:SCR_017097)

    Results from OddPub: We did not detect open data. We also did not detect open code. Researchers are encouraged to share open data when possible (see Nature blog).


    Results from LimitationRecognizer: We detected the following sentences addressing limitations in the study:
    The present study has some limitations. First, although we evaluate two predictions in the first week, evaluations might be different for a longer time or a different phase of the outbreak. Therefore, we must continue to evaluate the two predictions over a longer time. Secondly, the epidemic curve predicted by our model was not fixed until one month later. Therefore, evaluation for the prediction of our model might change up to one month. The obtained result should be regarded as tentative. By contrast, the evaluation of Google’s prediction will not change.

    Results from TrialIdentifier: No clinical trial numbers were referenced.


    Results from Barzooka: We did not find any issues relating to the usage of bar graphs.


    Results from JetFighter: We did not find any issues relating to colormaps.


    Results from rtransparent:
    • Thank you for including a conflict of interest statement. Authors are encouraged to include this statement when submitting to a journal.
    • Thank you for including a funding statement. Authors are encouraged to include this statement when submitting to a journal.
    • No protocol registration statement was detected.

    About SciScore

    SciScore is an automated tool that is designed to assist expert reviewers by finding and presenting formulaic information scattered throughout a paper in a standard, easy to digest format. SciScore checks for the presence and correctness of RRIDs (research resource identifiers), and for rigor criteria such as sex and investigator blinding. For details on the theoretical underpinning of rigor criteria and the tools shown here, including references cited, please follow this link.