Predicting disease progress with imprecise lab test results

作者：

Highlights：

• An “IR” loss and the corresponding learning method are proposed to deal with the imprecision range problem of lab test results.

• The proposed method can provide more accurate and stable prediction result for different kinds of task and diverse learning methods.

• The proposed method can be readily incorporated to models by considering timestamps of the input sequence.

摘要

•An “IR” loss and the corresponding learning method are proposed to deal with the imprecision range problem of lab test results.•The proposed method can provide more accurate and stable prediction result for different kinds of task and diverse learning methods.•The proposed method can be readily incorporated to models by considering timestamps of the input sequence.

论文关键词：Prediction,Neural networks,Imprecise data,Health care

论文评审过程：Received 11 December 2021, Revised 16 May 2022, Accepted 28 July 2022, Available online 30 August 2022.

论文官网地址：https://doi.org/10.1016/j.artmed.2022.102373