A likelihood-based convolution approach to estimate major health events in longitudinal health records data: an external validation study
Author Lisiane Pruinelli, PhD, MS, RN, FAMIA, discusses this month's JAMIA Journal Club selection:
Pruinelli L, Zhou J, Stai B, et al. A likelihood-based convolution approach to estimate major health events in longitudinal health records data: an external validation study. J Am Med Inform Assoc. 2021;28(9):1885-1891. doi:10.1093/jamia/ocab087.
We are now using Zoom for our JAMIA Journal Club webinars.
Registering may prompt you to create a Zoom account if you don’t already have one.
All JAMIA Journal Club Webinars are open access and free to the informatics community.
Lisiane Pruinelli PhD, MS, RN, FAMIA, is an Assistant Professor in the University of Minnesota School of Nursing. She earned a PhD degree from the University of Minnesota School of Nursing and a Masters of Sciences and a Bachelor’s of Nursing Sciences degrees from the Federal University of Rio Grande do Sul, Brazil. She teaches health informatics and data science for undergraduate and graduate students. In addition, she is an Affiliate Faculty at the Institute for Health Informatics, where she collaborates in research and advising.
Her area of research is on applying innovative nursing informatics tools and cutting-edge data science methods to investigate complex disease conditions. Her research goal is to identify the problems and target interventions that increase the quality of health. Her major research is in liver transplantation, but she has also developed methodologies for sepsis and pain management, using the same data science approach.
She is a Fellow of the American Medical Informatics Association. Previously she served as a co-chair for the MNRS. Currently, she serves as the co-chair of the NKBDS Initiative, co-chair for the Data Science and Clinical Analytics workgroup, and as an advisor board member for the IMIA-SEP, with more than 30 countries represented.
Statement of Purpose
Electronic Health Records (EHR) data present diagnosis codes commonly dated to admission or discharge dates, thus inferring the onset (index) date is required. We have combined related laboratory results, procedures, and/or medication prescriptions/administrations to infer the onset date of problems. Similarly, not all liver transplant (LT) procedures have dates. Although procedures typically have their dates recorded in all datasets, LT has its own ICD billing codes, hence recording the individual procedure codes (which would have specific dates) for billing is not necessary. To estimate the LT date, we developed a novel methodology that synthesizes information from multiple sources (labs, medications, procedures, billing and claims diagnoses) using a combined method derived from the field of signal processing. We were able to estimate the LT date with great precision, 92% of the estimates fell into our target window. We further externally validated this method in an unseen dataset with similar results.
The benefit of the proposed method goes beyond just completing specific missing data elements in multi-site datasets, but shows the potential for transferability. Missing time stamps are common even in single-site EHR data. For instance, specific dates of diagnoses are mostly not available due to the structure of the EHR: diagnosis codes are dated to the discharge and/or admission date as they are primarily used for billing. Similarly, during aggregation, data elements could be omitted due to lack of standards, interoperability, or a lack of documentation. The proposed method can be applicable to other health conditions beyond LT, specifically when estimating a missing event date that has related measurements and interventions with known dates.
The target audience for this activity is professionals and students interested in health informatics.
After participating in the webinar the learner should be better able to:
- Consider the importance of having exact time stamps in the EHR for longitudinal research;
- Review a novel convolution-based change detection methodology developed, tested and externally validated; and
- Consider transferring this method of estimating time events in the EHR for multiple major health events
- 35-minute presentation by article author(s) considering salient features of the published study and its potential impact on practice
- 25-minute discussion of questions submitted by listeners via the webinar tools and moderated by JAMIA Student Editorial Board members.
The American Medical Informatics Association is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians.
Credit Designation Statement
The American Medical Informatics Association designates this live activity for a maximum of 1 AMA PRA Category 1 Credit(s)™. Physicians should claim only the credit commensurate with the extent of their participation in the activity.
The live webinar only offers CME credit. The recording on our website will be openly available for learners but will not offer CME credit.
No commercial support was received for this activity.
Disclosures for this Activity
The following planners, and staff who are in a position to control the content of this activity disclose that they have no relevant financial relationships with commercial interests/ineligible entities:
Presenter: Lisiane Pruinelli
JAMIA Journal Club planners: Hannah Burkhardt, Kirk E. Roberts
AMIA staff: Susanne Arnold, Pesha Rubinstein
Instructions for Claiming CME Credit
Use the link in the webinar’s chat area to access the claim-credit survey; in a day or two you will receive an email with your CME certificate.
If you require a certificate of participation, contact Pesha@amia.org.