MSGID: 1:317/3 63f6ebdf   
   PID: hpt/lnx 1.9.0-cur 2019-01-08   
   TID: hpt/lnx 1.9.0-cur 2019-01-08   
    Research team creates statistical model to predict COVID-19 resistance   
    Proof-of-concept study shows promise for machine-learning system that   
   uses electronic health data to make its predictions    
      
     Date:   
         February 22, 2023   
     Source:   
         Johns Hopkins Medicine   
     Summary:   
         Researchers have created and preliminarily tested what they believe   
         may be one of the first models for predicting who has the highest   
         probability of being resistant to COVID-19 in spite of exposure   
         to SARS-CoV-2, the virus that causes it.   
      
      
         Facebook Twitter Pinterest LinkedIN Email   
   FULL STORY   
   ==========================================================================   
   Researchers from Johns Hopkins Medicine and The Johns Hopkins University   
   have created and preliminarily tested what they believe may be one of   
   the first models for predicting who has the highest probability of being   
   resistant to COVID-19 in spite of exposure to SARS-CoV-2, the virus that   
   causes it.   
      
      
   ==========================================================================   
   The study is reported online today in the journal PLOS ONE.   
      
   "If we can identify which people are naturally able to avoid infection   
   by SARS- CoV-2, we may be able to learn -- in addition to societal   
   and behavioral factors -- which genetic and environmental differences   
   influence their defense against the virus," says lead study author   
   Karen (Kai-Wen) Yang, a biomedical engineering graduate student in the   
   Translational Informatics Research and Innovation Lab at The Johns Hopkins   
   University. "That insight could lead to new preventive measures and more   
   highly targeted treatments."  For its study, the research team set out   
   to determine if a machine-learning statistical model could use health   
   characteristics stored in electronic health records -- providing patient   
   data such as comorbidities (other medical conditions) and prescribed   
   medications -- as a means to pinpoint people with a natural ability   
   to avoid SARS-CoV-2 infection. Those persons, says Yang, could then be   
   studied to better understand the factors enabling their resistance.   
      
   A machine-learning model is a computer program or system that uses   
   mathematical algorithms to find statistical patterns, and then apply the   
   patterns moving forward. This gives such systems the ability to imitate   
   human thinking and reasoning, and similar to the brain, learn over time.   
      
   "Using a machine-learning system to recognize complex patterns in large   
   numbers of people with COVID-19 enabled another team of Johns Hopkins   
   Medicine researchers in 2021 to predict the course of an individual   
   patient's case and determine the likelihood that it would become severe,"   
   says co-senior study author Stuart Ray, M.D., vice chair of medicine for   
   data integrity and analytics, and professor of medicine at the Johns   
   Hopkins University School of Medicine. "Based on their success, our   
   team wondered if the same approach also might be applied to predicting   
   who could be exposed to SARS-CoV-2 in close quarters and still not   
   get infected."  To demonstrate the model's ability to predict COVID-19   
   resistance, the researchers first acquired data from a clinical registry   
   called the Johns Hopkins COVID-19 Precision Medicine Analytics Platform   
   Registry (JH-CROWN). The registry contains information for patients   
   seen within the Johns Hopkins Health System who have been suspected of,   
   or confirmed as, having a SARS-CoV- 2 infection.   
      
   For their resistance study, the researchers only included individuals who   
   received a COVID-19 test between June 10, 2020, and Dec. 15, 2020, and   
   who reported "potential exposure to the virus" as the reason for testing.   
      
   The ending date was the point at which large-scale COVID-19 vaccination   
   efforts started in the United States. Choosing this date, the researchers   
   say, enabled them to avoid the effects on their findings of vaccines   
   preventing infection rather than natural resistance.   
      
   The 8,536 study participants who reported exposure as their reason   
   for getting COVID tested were divided into two groups: those who did   
   not share a residence (called a "household" in this study) with any   
   COVID-19 patients or their residence had 10 or more patients; and those   
   who shared a residence with 10 or fewer people, with at least one being   
   a COVID-19 patient. The first group, with 8,476 of the participants,   
   was designated as the Training and Testing Set, while the second group,   
   called the Household Index (HHI) Set, had 60 members, and was used as   
   a separate testing set.   
      
   Keeping the household number to 10 or fewer, the researchers say,   
   excluded people living in apartment complexes, dormitories and other   
   higher-density, multi-unit living areas where exposure to a particular   
   person positive for SARS-CoV-2 would be less intense.   
      
   To identify patterns and cluster participants so that those naturally   
   resistant to SARS-CoV-2 stand out, both study sets were analyzed using the   
   Maximal- frequent All-confident pattern Selection Pattern-based Clustering   
   (MASPC) algorithm. MASPC is specifically designed for electronic health   
   record data analysis that combines patient demographic information (age,   
   sex and race), the International Statistical Classification of Diseases   
   and Related Health Problems (ICD) medical diagnostic codes relevant to   
   each case, outpatient medication orders and the number of comorbidities   
   (other diseases) present.   
      
   "We hypothesized that MASPC would enable us to cluster patients   
   with similar patterns in their data to define them as resistant and   
   non-resistant to SARS- CoV-2, and with the hope that the algorithm would   
   learn with each analysis how to improve the accuracy and reliability of   
   future assignments," says Ray. "This initial study using JH-CROWN data   
   was conducted to give life to that hypothesis, a proof-of-concept trial   
   of our statistical model to show that resistance to COVID-19 might be   
   predictable based a patient's clinical and demographic profile."  "In the   
   Training and Testing Set, we identified 56 patterns of ICD codes split   
   into two groups: associated with resistance or not associated," Yang says.   
      
   "Statistical analyses of how well these patterns differentiated between   
   resistance and non-resistance yielded five patterns that did the best   
   job for our small and localized [Baltimore-Washington, D.C., metroplex]   
   study population to define who was most likely exposed to SARS-CoV-2."   
   "Looking for these patterns in HHI Set -- the individuals most likely   
   to have been exposed to SARS-CoV-2 in close quarters -- and then   
   statistically analyzing the results, our model's best performance was   
   0.61," says Ray. "Since a score of 0.5 shows only chance association   
   between the prediction and reality, and 1 is 100% association, this shows   
   the model has promise as a tool for identifying people with COVID-19   
   resistance who can be further studied," says Ray.   
      
   Limitations to the study, says Ray, include potential bias from   
   self-reporting of COVID-19 exposure by participants, the small number   
   of participants in the HHI group, the possibility that participants   
   tested for SARS-CoV-2 using home kits or at facilities outside the   
   Johns Hopkins system (and therefore, the tests were not recorded in the   
   JH-CROWN database), and the short timeframe of the study itself. He adds   
   that future trails using national patient data are needed to validate   
   the model's ability.   
      
   Along with Yang and Ray, the members of the study team from Johns Hopkins   
   Medicine and Johns Hopkins University are graduate and undergraduate   
   students Yijia Chen, Jacob Desman, Kevin Gorman, Chloe' Paris, Ilia   
   Rattsev, Tony Wei and Rebecca Yoo; and faculty co-senior authors Joseph   
   Greenstein and Casey Overby Taylor.   
      
   The study authors report no conflicts of interest.   
      
       * RELATED_TOPICS   
             o Health_&_Medicine   
                   # Personalized_Medicine # Today's_Healthcare #   
                   Diseases_and_Conditions # Diabetes   
             o Computers_&_Math   
                   # Computer_Modeling # Statistics # Hacking #   
                   Mathematical_Modeling   
       * RELATED_TERMS   
             o Severe_acute_respiratory_syndrome o Probability_distribution   
             o Virus o Uniform_distribution_(continuous) o Computer_virus   
             o Pandemic o Rabies o Epstein-Barr_virus   
      
   ==========================================================================   
   Story Source: Materials provided by Johns_Hopkins_Medicine. Note:   
   Content may be edited for style and length.   
      
      
   ==========================================================================   
   Journal Reference:   
      1. Kai-Wen K. Yang, Chloe' F. Paris, Kevin T. Gorman, Ilia Rattsev,   
      Rebecca   
         H. Yoo, Yijia Chen, Jacob M. Desman, Tony Y. Wei, Joseph   
         L. Greenstein, Casey Overby Taylor, Stuart C. Ray. Factors   
         associated with resistance to SARS-CoV-2 infection discovered using   
         large-scale medical record data and machine learning. PLOS ONE,   
         2023; 18 (2): e0278466 DOI: 10.1371/ journal.pone.0278466   
   ==========================================================================   
      
   Link to news story:   
   https://www.sciencedaily.com/releases/2023/02/230222210542.htm   
      
   --- up 51 weeks, 2 days, 10 hours, 50 minutes   
    * Origin: -=> Castle Rock BBS <=- Now Husky HPT Powered! (1:317/3)   
   SEEN-BY: 15/0 106/201 114/705 123/120 153/7715 226/30 227/114 229/111   
   SEEN-BY: 229/112 113 307 317 400 426 428 470 664 700 292/854 298/25   
   SEEN-BY: 305/3 317/3 320/219 396/45   
   PATH: 317/3 229/426