We’re hiring, join our Corporate Team, Information Technology, Finance, Clinical Research, and more! See all open opportunities

Data-Driven Kidney Care

Sep 19, 2022

Data Science has emerged in the medical industry as a tool that provides healthcare systems with multiple integrated approaches to data. These integrated approaches include data cleaning, storage, analysis, processing, interpretation, and the collection of large volumes of data. Examples of large volumes of data in kidney care include that generated in treating chronic kidney disease (CKD) and end-stage renal disease (ESRD). 

Healthcare is rapidly becoming data-dependent to deliver coordinated care to patients. The data-dependency has seen a shift of data leveraged from whole populations based on randomized controlled trials to personal data and specific databases. Thus coordinated care has developed into the delivery of customized individual patient treatment options. 

CKD incidence is increasing worldwide, with over 37 million Americans estimated to have CKD. The individual and societal costs of care are well documented. Therefore, the importance of the health management of chronic kidney disease has become central for governments and shareholders. 

Key Facts: 

  • Nephrology analytics enables the delivery of innovative data-driven, value-based kidney care. 
  • The usage of EHRs and their wide applications have created unique and novel opportunities for nephrology research. 
  • Progress in big data is driving the expanding implementation of artificial intelligence in kidney care.  

How can Data be Utilized to Drive Kidney Care?


Artificial intelligence (AI) is a computer algorithm designed to augment and mimic human actions and thought patterns. AI has been implemented in the healthcare industry due to the improved big-data storage devices and the adoption of electronic healthcare records (EHRs). AI has various roles in healthcare, not limited to medical data management, personalized treatment, drug development, health monitoring, and disease diagnosis. 

Machine learning (ML) is a subset of AI that allows computers to perform specific tasks without being issued detailed instructions. When used in predictive modeling (such as predicting ESKD in early-stage CKD patients), ML algorithms can be trained to capture underlying patterns of sample data, such as EHRs, pathology reports, and more. Then, algorithms make predictions about new data based on the information gathered. ML represents a more sophisticated math function compared to traditional statistics. 

Electronic Health Records (EHRs) contain patients’ medical histories and facilitate medical data processing using computers. EHRs are widely utilized in clinical informatics strategies. 

Transforming Kidney Care with Artificial Intelligence 

Artificial intelligence (AI), Automation, and Machine learning (ML) can provide a solution to help free up limited human resources that instead can be diverted and focused more on other areas of research, such as the slowing disease progression of renal disease. 

The goal of AI is not to replace physicians but to help improve and clarify physicians’ clinical decisions. Practical uses for AI in kidney care include diagnosis, medical image analysis, prognosis, and risk prediction. For example, in the diagnosis of CKD, Artificial Intelligence can help identify patients at risk and enhance the accuracy of the final diagnosis. AI can also help physicians precisely predict prognosis. 

Using Machine Learning to Improve Patient Outcomes Through Risk Indicators Assessment For CKD

Early stages of CKD have no prevalent symptoms and are often diagnosed during the later stages of the disease. Progressive loss of kidney function is indicative of end-stage kidney disease (ESKD), increasing the need for potential kidney replacement therapy. Timely intervention for patients at high risk of ESKD may improve the patient’s quality of life by delaying the disease progression and reducing morbidity and healthcare costs. 

According to a 2022 study, implementing a reliable prediction model for the risk of end-stage kidney disease during the early stages of CKD can prove to be clinically indispensable. These machine learning models also provide the data needed for physicians to implement customized individual treatment options and management for high-risk patients. 

In this study, statistical models were developed to predict the risk of an early stage CKD developing ESKD, with variables such as lab results, gender, age, race, albuminuria, and glomerular filtration rate (eGFR). In different studies, ML models were performed to predict the progress of CKD, leveraging data to calculate future eGRF values and estimate the risk of short-term mortality following dialysis. 

These models can be of clinical significance as urine tests are a critical diagnostic approach for CKD. Specifically, the level of albuminuria is regarded as a significant predictor for CKD disease progression. 

Data Science for Easy Follow-up of Long-Term Treatments

Data science can be valuable in identifying understandable patterns from collected data. A general data science approach to extracting relevant decisional knowledge from data, known as clustering based on rules by states (ClbRxS), helps support complex decision-making over time. 

The final outcome of ClbRxS is trajectory maps, which consist of a visual diagram displaying the most frequent paths of patients (objects) through the classes/levels of successive data waves. Trajectory maps are intuitive and provide an interpretable perspective of how patients progress over time. This allows healthcare experts to analyze the primary evolution patterns of individual patients and allocate appropriate treatments or actions. 

From a clinical decision point of view, ClbRxS is crucial to bridging the gap between the effective decision-making layer and the results of the data mining step. This methodology (clustering based on rules by states) has significant potential to assess the efficacy of long-term treatments through follow-ups of CKD patients. 

Innovations in Kidney Care Using Data-Driven Clinical Pathways


In terms of CKD, data-driven clinical pathway learning can be used to summarize longitudinal and multidimensional information gathered from EHRs into groups/clusters of prevalent sequences. For example, data such as the number of new or recurring CKD patient visits or the number of CKD patients hospitalized for complications can be collected. Once obtained, this data can be leveraged to review current practices and identify potential areas in the care delivery process that need adjusting or innovations. 

Panoramic Health: Data-Driven Kidney Care


Panoramic Health is an industry leader in data and predictive analytics, with the largest, real-time CKD database that provides episodic and longitudinal data from over 630,000 patients. The CKD database enables superior risk stratification and analytics that powers predictive analytics and data-driven care interventions, providing patients with holistic kidney care. The utilization of data and predictive analytics also plays an essential role in Panoramic Health’s comprehensive value-based care model that equips physicians and providers with a “plug-and-play” model. 

Patients can receive customized and individualized treatments based on their risk profile and disease stage. This may result in reduced hospitalizations, avoided crash dialysis, the delay or prevention of ESRD, and accelerated access to transplant evaluation or placement. 



There are many novel approaches to be designed, with the opportunities for AI, machine learning, and data science to slow disease progression in nephrology. While some challenges remain, data-driven care plays a crucial role in improving patient outcomes and innovating the general practice of nephrology.