
5. Clinical Medicine
We use developed mathematical model to predict severity of infection and compare different treatment.
We employ our developed mathematical models to predict the severity of infections by analyzing factors such as viral load dynamics, immune response, and patient-specific characteristics. These models help identify high-risk individuals, estimate disease progression, and assess potential complications. Additionally, we use them to compare the effectiveness of different treatment strategies, evaluating factors like drug efficacy, optimal dosing regimens, and resistance development. By integrating clinical and experimental data, our approach supports personalized treatment decisions, enhances therapeutic strategies, and informs clinical guidelines for better patient outcomes.
As an example, we analyze clinical data using the viral dynamics model to identify patterns with severe infections.
Regarding the association between viral load and clinical outcomes such as mortality and clinical scores, a number of studies suggested that a high viral load at diagnosis is associated with more severe clinical outcomes and increased mortality. Such differences in viral load may reflect differences in the immunological response. For example, some studies suggest that severity is associated with the production of autoantibodies, which impairs innate and intrinsic antiviral immunological responses. Autoantibodies perturb immune function by neutralizing the cytokines/chemokines, thus could be associated with disease severity. Autoantibodies may also be associated with persistent viral load for a longer time. Thus to associate viral load and clinical outcomes, temporal viral load pattern should be considered, rather than single time pint viral loads.
Here, we propose a classification model to predict future severity/mortality risk using longitudinal viral load data collected during the early phase of the infection. Patients will be categorized into several groups using the estimated viral load data with a weight function. The weight function will be optimized to realize the best predictive and discriminative ability of mortality (and other clinical outcomes such as critical disease). This project will identify the critical patterns of viral dynamics and the critical timing of viral load collection to identify vulnerable patients.
Related publications:
Ejima K, Xavier NA, Mehta T* (2020). Comparing the ability of two comprehensive clinical staging systems to predict mortality: EOSS and CMDS. Obesity 28:353-361
Howell CR*, Mehta T, Ejima K, Ness KK, Cherrington A, Fontaine KR (2018). Body composition and mortality in Mexican American adults: results from the National Health and Nutrition Examination Survey. Obesity 26:1372-80
Zuidema TR, Bazarian JJ, Kercher KA, Mannix R, Kraft RH, Newman SD, Ejima K, Rettke DJ, Macy JT, Steinfeldt JA, Kawata K* (2023). Longitudinal Associations of Clinical and Biochemical Head Injury Biomarkers With Head Impact Exposure in Adolescent Football Players. JAMA Netw Open. 6(5):e2316601
Nowak MK, Bevilacqua ZW, Ejima K, Huibregtse ME, Chen Z, Mickleborough TD, Newman SD, Kawata K* (2020). Neuro-ophthalmologic response to repetitive subconcussive head Impacts: A randomized clinical trial. JAMA Ophthalmology 138:350-357