Why do we need data analytics tools in laboratory medicine?

As clinical laboratorians, we rely heavily on metrics derived from laboratory data to assure the best quality of laboratory results delivered to support patient care. Clinical laboratories in the new era are equipped with high-throughput, high-volume automated lines and complicated assays with cutting-edge technology. Massive amounts of laboratory data are generated rapidly and with increasingly higher complexity. The rise of ‘Big Data’ means all fields are becoming “data science” fields, resulting in major data processing challenges for laboratorians. Advanced data analysis tools, many of which are open source, are available and have been successfully applied in fields other than healthcare. We hope this article will help our fellow clinical chemists to understand the capabilities of popular data analysis tools, realize the impact of application of those tools in lab medicine, and eventually enable selection of appropriate tools for efficient data analysis to serve the needs of the clinical lab.

Data analytics applications in laboratory medicine

Data analytics tools are utilized by laboratorians to support quality assurance activities, lab management/operations, and clinical applications (Table 1). All clinical labs are engaged in the quality assurance process as an essential part of regulatory compliance (as per ISO 15189:2012) in order to ensure control throughout the total analytical process (pre-analytical, analytical, and post-analytical), identify errors and facilitate interventions for continuous improvement. Data analytics tools can be leveraged to automate analysis of millions of data points for prompt quality assurance including monitoring of quality indicators, such as pre-analytical error rates, critical result call times, turnaround times and numerous other metrics such as those developed by the International Federation of Clinical Chemistry (IFCC) Working Group on “Laboratory Errors and Patient Safety'' [1]. This can be accomplished through development of automated reports- either periodically or real-time. Additionally, the use of interactive dashboards allows visual display and enables the data-driven decision-making process.

Data analytics tools have also been widely used for various clinical applications. Using data obtained from the Laboratory Information System (LIS) and/or Electronic Health Records (EHR), we are able to easily establish and verify direct or indirect reference intervals and diagnostic thresholds [2] . Use of data analytics tools can be particularly helpful for this purpose when probing large datasets, interrogating multiple tests, or for tests influenced by co-variates such as age and sex. As another clinical application, machine learning tools can be used to improve post-analytical throughput, automated and support high-complexity test result interpretation, and increase reliability and reproducibility of interpretations that have been traditionally subjective when performed by humans. Interpretation of serum protein electrophoresis results or plasma amino acid profiles are two notable examples [3,4]. While the above applications have been typically applied locally to a single healthcare institution or across a healthcare system, a major opportunity for lab data analytics exists in harnessing big data across labs to support patient care as patients move throughout the healthcare system with numerous potential clinical aFloris Chabrun et al. “Achieving Expert-Level Interpretation of Serum Protein Electrophoresis through Deep Learning Driven by Human Reasoning.” Clinical Chemistry, Volume 67, Issue 10, October 2021, Pages 1406–1414.pplications including data mining and evidence-based harmonization of lab test reporting to facilitate optimal test result interpretation [5].

What are the specific tools and learning resources?

Data analytics tools can range from those used for data processing, visualization, quantitative analysis, and machine learning. Open-source data analytics tools such as the R program and Python require one to develop programming skills, and correspondingly, there would be a learning curve. To ease this learning process, there are numerous readily available resources, such as courses offered at national conferences and online forums (see supplemental document for a list of learning resources below). Additionally, there are a variety of user-friendly tools not requiring advanced programming skills such as Microsoft Excel and Power BI, GraphPad, Tableau, and tools specific to one’s EHR such as SlicerDicer that could be utilized for a variety of data analyses, reporting, and visualization purposes. Lastly, laboratories should increase the collaboration with information technology, applications development, and business intelligence teams which are becoming increasingly supported in healthcare and could be a valuable resource to help meet the data analytics needs of the clinical lab.

Acknowledgement

We thank Drs. Thomas Durant (Yale University) and Dustin Bunch (Nationwide Children’s Hospital) for contribution to the learning resources.

References

  1. Sciacovelli L et al; “Quality Indicators in Laboratory Medicine: the status of the progress of IFCC Working Group "Laboratory Errors and Patient Safety" project.” Clin Chem Lab Med. 2017 Mar 1;55(3):348-357.
  2. Bunch DR. Indirect reference intervals using an R pipeline. J Mass Spectrom Adv Clin Lab. 2022 Feb 23;24:22-30
  3. Floris Chabrun et al. “Achieving Expert-Level Interpretation of Serum Protein Electrophoresis through Deep Learning Driven by Human Reasoning.” Clinical Chemistry, Volume 67, Issue 10, October 2021, Pages 1406–1414.
  4. Wilkes, et al. "A machine learning approach for the automated interpretation of plasma amino acid profiles." Clinical Chemistry 66.9 (2020): 1210-1218.
  5. Parker, M.L., Adeli, Khosrow and on behalf of the CSCC Working Group on Reference Interval Harmonization,. "Pediatric and adult reference interval harmonization in Canada: an update" Clinical Chemistry and Laboratory Medicine (CCLM), vol. 57, no. 1, 2019, pp. 57-60.
  6. Herman DS et al, Artificial Intelligence and Mapping a New Direction in Laboratory Medicine: A Review. Clin Chem. 2021 Nov 1;67(11):1466-1482.

 

Table 1: Applications for data analytics tools

Table 1: Applications for data analytics tools