Abstract
Malnutrition is considered a pressing global health concern because of its role as a contributing factor for almost half of all childhood deaths (Walson & Berkley, 2018) and as an agent in developmental impediments. Gut dysfunction and altered gut microbiota have been linked to clinical outcomes of infantile malnutrition (Subramanian et al., 2014). Thus, nutritional rehabilitation therapies have been developed to help restore a healthy gut microbiome. However, current therapies have proved to be ineffective for sustainable growth. A promising new strategy is the transfer of live gut microbes to restore the gut microbiome. To become feasible for large-scale administration, new strategies must be employed to increase manufacturing yield of these human gut bacteria (O’Toole, Marchesi, & Hill, 2017). This project aims to use optimization techniques such as flux variability analysis (FVA) and parsimonious flux balance analysis (pFBA) to simulate nutrient preferences using genome scale metabolic network models from gut microbes. We developed a computational pipeline involving an iterative process of pFBA that can be applied to various probiotic strains to develop genome scale metabolic network models (GENREs) and nutrient preferences. These nutrient preferences were validated by experimental Biolog data by determining the correlation between the nutrient ranks from the pFBA analysis and experimental results. Nutrient preferences were simulated for a multispecies probiotic to look at the results in context.
Most Americans will receive a medical misdiagnosis at least once in their lifetime (Rue, 2019). Medical misdiagnoses occur in 5% of outpatient office visits, 10% of hospital inpatient deaths, and 12% of hospital adverse events, and contribute to 74,000 deaths per year (Papier, 2015). Artificial intelligence (AI) will be utilized to help with diagnosis of diseases by revealing previously hidden trends in data, and thus will have substantial impact both at the individual patient and system level (Panch et al., 2018). However, these systems prompt concerns related to the privacy of patient data, the quality and safety of these algorithms, and their impacts on the role of physicians and the healthcare system at large. This thesis seeks to answer where is the need for future policies surrounding artificial intelligence for diagnosis to protect patient privacy while providing the best possible care. Due to the risks associated with this technology, AI for the diagnosis of diseases must be considered an inherently political technology. This classification is in part due to how its development may conflict with current regulations and the need for government and healthcare systems to protect the safety and privacy of patients. This paper presents an analysis of the potential impacts of the use of artificial intelligence for the diagnosis of disease and the effects of current policies and regulations on its development and implementation. Additional policies are proposed for the use of this technology while focusing on the protection of the patients.
