Inactivated Vaccine Dosage and Serum IgG Levels Correlate with Persistent COVID-19 Infection in Hematologic Malignancy Patients During the Omicron Surge in China (preprint)

In hematologic malignancies (HM) patients, COVID-19 infections carry a significant risk of mortality due to disease status, treatment, and other factors.The risk factors of the severity and persistence of COVID-19 infections remains unclear. A study observed adults with HM diagnosed with COVID-19 from November 2022 to February 2023. Patient blood samples yielded biochemical data, with COVID-19 confirmed via RNA or antigen testing. In the examined cohort, 133 individuals diagnosed with HM and concomitantly infected with COVID-19 were scrutinized. Using advanced multivariate logistic regression, high C-reactive protein levels (≥100mg/L) significantly increased the risk of severe/critical conditions in HM patients with COVID-19 (OR: 3.415, 95% CI: 1.294-9.012; p=0.013). Patients enduring Omicron infection beyond 30 days were deemed persistent, in contrast to those achieving infection control within this duration. The research indicated that taking ＜2 vaccine doses (OR: 0.202, 95% CI: 0.048-0.857; p=0.030), having low IgG levels (＜1000 mg/dl) (OR: 0.129, 95% CI: 0.027-0.607; p=0.010), and increased interleukin-6 levels (≥12pg/ml) (OR: 5.098, 95% CI: 1.118-23.243; p=0.035) were key indicators of ongoing infection. A significant difference in survival rates was observed between patients with persistent and non-persistent infections, with the latter showing better survival outcomes (P<0.001). In conclusion, increased C-reactive protein levels had a higher likelihood of severe health outcomes for HM patients with COVID-19 infection. Persistent infections tended to be more prevalent in those with lower vaccine dosages, diminished IgG levels, and escalated interleukin-6 levels.

Advancing COVID-19 Diagnosis with Privacy-Preserving Collaboration in Artificial Intelligence (preprint)

Artificial intelligence (AI) provides a promising substitution for streamlining COVID-19 diagnoses. However, concerns surrounding security and trustworthiness impede the collection of large-scale representative medical data, posing a considerable challenge for training a well-generalised model in clinical practices. To address this, we launch the Unified CT-COVID AI Diagnostic Initiative (UCADI), where the AI model can be distributedly trained and independently executed at each host institution under a federated learning framework (FL) without data sharing. Here we show that our FL model outperformed all the local models by a large yield (test sensitivity /specificity in China: 0.973/0.951, in the UK: 0.730/0.942), achieving comparable performance with a panel of professional radiologists. We further evaluated the model on the hold-out (collected from another two hospitals leaving out the FL) and heterogeneous (acquired with contrast materials) data, provided visual explanations for decisions made by the model, and analysed the trade-offs between the model performance and the communication costs in the federated training process. Our study is based on 9,573 chest computed tomography scans (CTs) from 3,336 patients collected from 23 hospitals located in China and the UK. Collectively, our work advanced the prospects of utilising federated learning for privacy-preserving AI in digital health.

A collaborative online AI engine for CT-based COVID-19 diagnosis (preprint)

Artificial intelligence can potentially provide a substantial role in streamlining chest computed tomography (CT) diagnosis of COVID-19 patients. However, several critical hurdles have impeded the development of robust AI model, which include deficiency, isolation, and heterogeneity of CT data generated from diverse institutions. These bring about lack of generalization of AI model and therefore prevent it from applications in clinical practices. To overcome this, we proposed a federated learning-based Unified CT-COVID AI Diagnostic Initiative (UCADI, http://www.ai-ct-covid.team/), a decentralized architecture where the AI model is distributed to and executed at each host institution with the data sources or client ends for training and inferencing without sharing individual patient data. Specifically, we firstly developed an initial AI CT model based on data collected from three Tongji hospitals in Wuhan. After model evaluation, we found that the initial model can identify COVID from Tongji CT test data at near radiologist-level (97.5% sensitivity) but performed worse when it was tested on COVID cases from Wuhan Union Hospital (72% sensitivity), indicating a lack of model generalization. Next, we used the publicly available UCADI framework to build a federated model which integrated COVID CT cases from the Tongji hospitals and Wuhan Union hospital (WU) without transferring the WU data. The federated model not only performed similarly on Tongji test data but improved the detection sensitivity (98%) on WU test cases. The UCADI framework will allow participants worldwide to use and contribute to the model, to deliver a real-world, globally built and validated clinic CT-COVID AI tool. This effort directly supports the United Nations Sustainable Development Goals' number 3, Good Health and Well-Being, and allows sharing and transferring of knowledge to fight this devastating disease around the world.

Basic reproduction number of 2019 Novel Coronavirus Disease in Major Endemic Areas of China: A latent profile analysis (preprint)

Objective: The aim of the study is to analyze the latent class of basic reproduction number (R0) trend of 2019 novel coronavirus disease (COVID-19) in major endemic areas of China. Methods The provinces that reported more than 500 cases of COVID-19 till February 18, 2020 were selected as the major endemic area. The Verhulst model was used to fit the growth rate of cumulative confirmed cases. The R0 of COVID-19 was calculated using the parameters of severe acute respiratory syndrome (SARS) and COVID-19, respectively. The latent class of R0 was analyzed using a latent profile analysis model. Results The median R0 calculated from SARS and COVID-19 parameters were 1.84 - 3.18 and 1.74 - 2.91, respectively. The R0 calculated from the SARS parameters was greater than that of calculated from the COVID-19 parameters (Z = -4.782 - -4.623, P < 0.01). Both R0 can be divided into three latent classes. The initial value of R0 in class 1 (Shandong Province, Sichuan Province and Chongqing Municipality) was relatively low and decreases slowly. The initial value of R0 in class 2 (Anhui Province, Hunan Province, Jiangxi Province, Henan Province, Zhejiang Province, Guangdong Province and Jiangsu Province) was relatively high and decreases rapidly. Moreover, the initial value of R0 of class 3 (Hubei Province) was between that of class 1 and class 2, but the higher level of R0 lasts longer and decreases slowly. Conclusion The results indicated that overall trend of R0 has been falling with the strengthening of China's comprehensive prevention and control measures for COVID-19, however, presents regional differences.

Analysis on epidemic situation and spatiotemporal changes of COVID-19 in Anhui / 中华预防医学杂志

We used the epidemic data of COVID-19 published on the official website of the municipal health commission in Anhui province. We mapped the spatiotemporal changes of confirmed cases, fitted the epidemic situation by the population growth curve at different stages and took statistical description and analysis of the epidemic situation in Anhui province. It was found that the cumulative incidence of COVID-19 was 156/100 000 by February 18, 2020 and the trend of COVID-19 epidemic declined after February 7, changing from J curve to S curve. The actual number of new cases began to decrease from February 2 to February 4 due to the time of case report and actual onset delayed by 3 to 5 days.