Prognostic Model of COVID-19 Severity and Survival among Hospitalized Patients Using Machine Learning Techniques.

We conducted a statistical study and developed a machine learning model to triage COVID-19 patients affected during the height of the COVID-19 pandemic in Hong Kong based on their medical records and test results (features) collected during their hospitalization. The correlation between the values of these features is studied against discharge status and disease severity as a preliminary step to identify those features with a more pronounced effect on the patient outcome. Once identified, they constitute the inputs of four machine learning models, Decision Tree, Random Forest, Gradient and RUSBoosting, which predict both the Mortality and Severity associated with the disease. We test the accuracy of the models when the number of input features is varied, demonstrating their stability; i.e., the models are already highly predictive when run over a core set of (6) features. We show that Random Forest and Gradient Boosting classifiers are highly accurate in predicting patients' Mortality (average accuracy ∼99%) as well as categorize patients (average accuracy ∼91%) into four distinct risk classes (Severity of COVID-19 infection). Our methodical and broad approach combines statistical insights with various machine learning models, which paves the way forward in the AI-assisted triage and prognosis of COVID-19 cases, which is potentially generalizable to other seasonal flus.

Evaluation of the commercially available LightMix® Modular E-gene kit using clinical and proficiency testing specimens for SARS-CoV-2 detection.

BACKGROUND: Rapid and sensitive diagnostic assays for SARS-CoV-2 detection are required for prompt patient management and infection control. The analytical and clinical performances of LightMix® Modular SARS and Wuhan CoV E-gene kit, a widely used commercial assay for SARS-CoV-2 detection, have not been well studied. OBJECTIVE: To evaluate the performance characteristics of the LightMix® E-gene kit in comparison with well-validated in-house developed COVID-19 RT-PCR assays. STUDY DESIGN: Serial dilutions of SARS-CoV-2 culture isolate extracts were used for analytical sensitivity evaluation. A total of 289 clinical specimens from 186 patients with suspected COVID-19 and 8 proficiency testing (PT) samples were used to evaluate the diagnostic performance of the LightMix® E-gene kit against in-house developed COVID-19-RdRp/Hel and COVID-19-N RT-PCR assays. RESULTS: The LightMix® E-gene kit had a limit of detection of 1.8 × 10-1 TCID50/mL, which was one log10 lower than those of the two in-house RT-PCR assays. The LightMix® E-gene kit (149/289 [51.6%]) had similar sensitivity as the in-house assays (144/289 [49.8%] for RdRp/Hel and 146/289 [50.5%] for N). All three assays gave correct results for all the PT samples. Cycle threshold (Cp) values of the LightMix® E-gene kit and in-house assays showed excellent correlation. Reproducibility of the Cp values was satisfactory with intra- and inter-assay coefficient of variation values <5%. Importantly, the LightMix® E-gene kit, when used as a stand-alone assay, was equally sensitive as testing algorithms using multiple COVID-19 RT-PCR assays. CONCLUSIONS: The LightMix® E-gene kit is a rapid and sensitive assay for SARS-CoV-2 detection. It has fewer verification requirements compared to laboratory-developed tests.

Development of a Novel, Genome Subtraction-Derived, SARS-CoV-2-Specific COVID-19-nsp2 Real-Time RT-PCR Assay and Its Evaluation Using Clinical Specimens.

The pandemic novel coronavirus infection, Coronavirus Disease 2019 (COVID-19), has affected at least 190 countries or territories, with 465,915 confirmed cases and 21,031 deaths. In a containment-based strategy, rapid, sensitive and specific testing is important in epidemiological control and clinical management. Using 96 SARS-CoV-2 and 104 non-SARS-CoV-2 coronavirus genomes and our in-house program, GolayMetaMiner, four specific regions longer than 50 nucleotides in the SARS-CoV-2 genome were identified. Primers were designed to target the longest and previously untargeted nsp2 region and optimized as a probe-free real-time reverse transcription-polymerase chain reaction (RT-PCR) assay. The new COVID-19-nsp2 assay had a limit of detection (LOD) of 1.8 TCID50/mL and did not amplify other human-pathogenic coronaviruses and respiratory viruses. Assay reproducibility in terms of cycle threshold (Cp) values was satisfactory, with the total imprecision (% CV) values well below 5%. Evaluation of the new assay using 59 clinical specimens from 14 confirmed cases showed 100% concordance with our previously developed COVID-19-RdRp/Hel reference assay. A rapid, sensitive, SARS-CoV-2-specific real-time RT-PCR assay, COVID-19-nsp2, was developed.