Date of Graduation


Document Type


Degree Type



Statler College of Engineering and Mineral Resources


Industrial and Managements Systems Engineering

Committee Chair

Imtiaz Ahmed

Committee Co-Chair

Mohammad Abdullah Al-Mamun

Committee Member

Mohammad Abdullah Al-Mamun

Committee Member

Ashish Nimbarte



Developing Artificial Intelligence tools to investigate the phenotypes and correlates of Chronic Kidney Disease patients in West Virginia

Marzieh Amiri Shahbazi

Chronic kidney disease (CKD) is responsible for disrupting the lives of 37 million people just in the USA, which is about 1 in 7 adults. CKD results in a gradual loss of kidney function over time. Sometimes CKD doesn’t produce any significant symptoms until it reaches an advanced stage. On the other hand, acute kidney injury (AKI) accounts for a sudden decline in the kidney’s function. As a result, the kidneys fail to filter waste materials from the blood and cause an increase in blood pressure. High blood pressure can cause heart disease and, in the long-term, induce CKD. Literature to date says AKI leads to long-term adverse kidney outcomes and linked to CKD. AKI diagnosis, its severity, treatment, and recovery process have a major impact on the likelihood of a future diagnosis of CKD. This research attempts to understand the patient’s trajectory toward developing CKD after AKI diagnosis, key triggers contributing to this trajectory and ultimately develop an Artificial intelligence-based prognosis tool. To comprehend the role of AKI and previous hospitalization in the progress of CKD, various cohorts of CKD patients are created: i) AKI after hospitalization before CKD ii) Random AKI before CKD, and iii) No AKI before CKD. Prior comorbidities, medications, lab results, and pertinent procedures are considered, and for each cohort of patients, the most prevalent phenotypes are identified. The patient cohorts required for this analysis are generated from CKD patients residing in West Virginia. The data is provided by TriNetx, a global network platform. K-means clustering, and the latent class analysis (LCA) approach is used to identify and group the phenotypes of CKD for each cohort. The high-risk patient groups generated by the clustering algorithms are compared with each other. These results will help clinicians to understand the risk factors of CKD and the overall trajectory of the development of CKD. This research suggests that a single method of care does not work for all patients since phenotypes vary for distinct groups of patients and categorizing patients into distinct groups allows for the allocation of different resources and strategies for the care of different groups of patients. From this research, it is evident that patients’ risk profiles change over the years before developing CKD. There are also similarities as well as differences across the cohorts for each year, which suggests that CKD risk factors may be linked to prior AKI, hospitalization, or inpatient care.

fillable-ETD-committee-signature-form__AN (1) (1).pdf (231 kB)
Completed thesis and dissertation signature form