Reliability-based cleaning of noisy training labels with inductive conformal prediction in multi-modal biomedical data mining (preprint)

Accurately labeling biomedical data presents a challenge. Traditional semi-supervised learning methods often under-utilize available unlabeled data. To address this, we propose a novel reliability-based training data cleaning method employing inductive conformal prediction (ICP). This method capitalizes on a small set of accurately labeled training data and leverages ICP-calculated reliability metrics to rectify mislabeled data and outliers within vast quantities of noisy training data. The efficacy of the method is validated across three classification tasks within distinct modalities: filtering drug-induced-liver-injury (DILI) literature with title and abstract, predicting ICU admission of COVID-19 patients through CT radiomics and electronic health records, and subtyping breast cancer using RNA-sequencing data. Varying levels of noise to the training labels were introduced through label permutation. Results show significant enhancements in classification performance: accuracy enhancement in 86 out of 96 DILI experiments (up to 11.4%), AUROC and AUPRC enhancements in all 48 COVID-19 experiments (up to 23.8% and 69.8%), and accuracy and macro-average F1 score improvements in 47 out of 48 RNA-sequencing experiments (up to 74.6% and 89.0%). Our method offers the potential to substantially boost classification performance in multi-modal biomedical machine learning tasks. Importantly, it accomplishes this without necessitating an excessive volume of meticulously curated training data.

Tri-light Warning System for Hospitalized COVID-19 Patients]{A Tri-light Warning System for Hospitalized COVID-19 Patients: Credibility-based Risk Stratification under Data Shift (preprint)

Early warning of the novel coronavirus pneumonia (COVID-19) during the evolving pandemic waves is crucial for the timely treatment of patients and optimization of medical resource allocation. However, prior AI-based models often lack the reliability and performance validation under data distribution drifts, and are therefore problematic to be reliably utilized in real-world clinical practice. To address this challenge, we developed a tri-light warning system based on conformal prediction for rapidly stratification of COVID-19 inpatients. This system can automatically extract radiomic features from CT images and integrate clinical record information to output a prediction probability, as well as a credibility of each prediction. This system classifies patients in the general ward into red label (high risk) indicating a possible admission to ICU care, yellow label (uncertain risk) indicating closer monitoring, and green label (low risk) indicating a stable condition. The subsequent health policies can be further designed based on this system according to the specific needs of different hospitals. Extensive experiment from a multi-center cohort (n= 8,721) shows that our method is applicable to both the original strain and the variant strains of COVID-19. Given the rapid mutation rate of COVID-19, the proposed system demonstrates its potential to identify epidemiological risks early to improve patient stratification performance under data shift.

CT-based Rapid Triage of COVID-19 Patients: Risk Prediction and Progression Estimation of ICU Admission, Mechanical Ventilation, and Death of Hospitalized Patients (preprint)

The wave of COVID-19 continues to overwhelm the medical resources, especially the stressed intensive care unit (ICU) capacity and the shortage of mechanical ventilation (MV). Here we performed CT-based analysis combined with electronic health records and clinical laboratory results on Cohort 1 (n = 1662 from 17 hospitals) with prognostic estimation for the rapid stratification of PCR confirmed COVID-19 patients. These models, validated on Cohort 2 (n = 700) and Cohort 3 (n = 662) constructed from 9 external hospitals, achieved satisfying performance for predicting ICU, MV and death of COVID-19 patients (AUROC 0.916, 0.919 and 0.853), even on events happened two days later after admission (AUROC 0.919, 0.943 and 0.856). Both clinical and image features showed complementary roles in events prediction and provided accurate estimates to the time of progression (p