Please use this identifier to cite or link to this item: http://hdl.handle.net/10609/146579
Title: Finite mixture models for trajectory analysis of in-hospital routine laboratory values: application as biomarkers in spinal cord injury patients
Author: Torres Espin, Abel
Tutor: Fernández Martínez, Daniel  
Abstract: Background: Early diagnostic and prognostication after acute traumatic spinal cord injury (SCI) is challenging due to pathology complexities and population heterogeneity. Routinely collected data during standard medical practice, such as laboratory analytes, can serve as surrogates of underlying pathophysiological processes and therefore be used as a biomarker. We hypothesized that distinct temporal trends of blood analytes can be modeled after SCI and that those would be predictive of patient characteristics. Methods: Using real-world data from available electronic health records, we assembled a big-data asset and modeled distinct laboratory analytes measured over time during the early hospitalization after acute spine trauma with or without SCI. We fitted longitudinal finite mixture models (FMM) to determine distinct group trajectories over time on 20 blood analytes commonly measured in these populations. The probability of group trajectory membership was used in machine learning models to predict patient characteristics. Results: We show non-linear heterogeneous temporal trends of blood analytes after spine trauma and SCI. These trajectories are associated with different patient characteristics. In dynamic prediction experiments, the probability of belonging to a specific analyte trajectory is predictive of whether the patient would die in hospital, the patient presented with an SCI, and SCI severity. Conclusions: Routinely real-world data can be used to model blood analytes' dynamic changes after SCI with prediction validity for patient characteristics. Our work suggests that temporal blood trends are promising early predictors of SCI pathology. This work sets the bases for further developing dynamic biomarkers in neurotrauma and other neurological conditions.
Keywords: finite mixture models
trajectory analysis
spinal cord injury
Document type: info:eu-repo/semantics/masterThesis
Issue Date: 19-Jun-2022
Publication license: http://creativecommons.org/licenses/by/3.0/es/  
Appears in Collections:Trabajos finales de carrera, trabajos de investigación, etc.

Files in This Item:
File Description SizeFormat 
atorresespFMDP0622report.pdfReport of TFM10,56 MBAdobe PDFThumbnail
View/Open
Share:
Export:
View statistics

This item is licensed under aCreative Commons License Creative Commons