EKGAnalysisTool

• 

  Brainstorming ideas for logical analysis

Inspiration

Better Heart health

What it does

Version 1.1 -Includes: File Conversion, viewing of EKGs, and a guide of how to manually check for abnormalities.

How we built it

Built it using IJulia and Gadfly in Julia

Challenges we ran into

Converting files in Julia, and properly constructing structures

Accomplishments that we're proud of

Converting files in Julia, and properly constructing structures

What we learned

We gained a deeper understanding of EKG analysis as well as a greater appreciation for our local clinicians and cardiologists

What's next for EKGAnalysisTool

Future release goals: -implement an automatic check to reduce work for the clinician -implement a comparison to a standard of community based EKG data -Enhance database logging using attributes to achieve highly specific EKG comparison

An Excerpt from an A&P supplemental instructor and Bio Major

Personalized medicine involves modifying healthcare practices to meet the needs of individual patients. The idea is that using norms and averages from very large populations will not lead to the best care, on an individual level. Rather than comparing to standards from very large populations, biomarkers or risk assessments can be used to assess the best course of treatment, based on data from other similar patients [1]. Big data has the potential to allow for advances in personalized medicine. There are difficulties however, as biological systems are inherently complex, and the amount of data can be very large. As such, it is very important that new approaches be explored and methods designed to harness the power of data systems to be used in patient care [2]. The goal of the current project was to build a system to manage electrocardiogram (EKG or ECG) data, to facilitate personalized medicine, while also giving clinicians a tool to help interpret EKG data. Clinicians who are not cardiologists may be lacking the experience necessary to detect possible important abnormalities on EKG tracings. In fact, there is evidence to suggest that when compared to cardiologists, general practitioners were far less accurate at detecting certain abnormalities on EKG tracings [3]. When used correctly, the EKG can be a very important tool to detect disease. As such, reading an EKG should be approached in a systematic manner [4]. We built the framework for a program that functions to collect and store electrocardiogram data. The idea is that this paradigm can then be used by clinics to build a repository of local EKG data categorized by demographics (such as age, race, sex, history of illness, etc.). We argue that this would help to facilitate personalized medical care, as EKG tracings for individual patients can be compared to tracings from other patients of similar demographics, as opposed to using textbook tracings as standards. There are individual variations in EKG tracings. For example, some healthy individuals may have extra T waves following the QRS complex. Furthermore, it is normal for individuals of Afro-Caribbean decent to have downward deflecting T waves, but for other individuals, a downward deflection may be indicative of ischemia or previous heart attack [5]. If EKG standards are developed at the local level, and norms are generated by use of data from individuals very similar to the patient, then it only makes sense that a more individually appropriate treatment plan would be the result. Using EKG diagnostic criteria explained Ashley and Niebuer [5], we developed a prototype of our envisioned program. The way it would work in practice, would be to have a local clinic generate EKG data for varying subsets of populations, by storing their EKG trace data, along with relevant demographic information. This would establish a set of norms for particular patient characteristics for the particular local area. This repository would then contain healthy controls (such as tracings from patients in for routine checkups) as well patients with known diseases (such as patient with previous histories of hearth attacks). In practice, after a sufficient repository of data had been gathered, a patient given an EKG would then have this data entered into the program we built. The program reminds/walks the clinician through a systematic reading of the EKG tracing, asking simple questions, such as Is the S-T segment less than half of the R-R segment? (this systematic process can be seen in our program). After the clinician has responded to the promptings of the program, any abnormalities would be flagged for further examination. Meanwhile, the program will identify waveforms on the EKG and make relevant comparisons to the EKG tracings from the repository for similar individuals from that locale, flagging anomalies. (Note that our prototype program is only able to identify the R and T waves and make one comparison involving the height of the T wave. We envision that future versions of the program would identify all waveforms and would be able to make more relevant comparisons to supplement the systematic walk-through for the clinician. We would also envision that the systematic walk through would be further developed with input from cardiologists.) We learned first-hand the difficult in applying data systems to biological systems, and as mentioned above, this is a common difficulty in designing data methods for use in personalized medicine. One major difficulty that we faced involved the identification of waveforms on the EKG. There was much variation in the EKG tracings that we used (we took EKG tracing from 30 willing student participants) to construct the logic of the program. The experimental tracings did not match perfectly to the textbook example, and thus we were forced to come up with creative ways to identify waveforms.

Full Citations in Order of Their Use: [1] Fröhlich, H., Balling, R., Beerenwinkel, N., Kohlbacher, O., Kumar, S., Lengauer, T., ... & Zupan, B. (2018). From hype to reality: data science enabling personalized medicine. BMC medicine, 16(1), 1-15. [2] Cirillo, D., & Valencia, A. (2019). Big data analytics for personalized medicine. Current opinion in biotechnology, 58, 161-167. https://doi.org/10.1016/j.copbio.2019.03.004 [3] Compiet, S. A. M., Willemsen, R. T. A., Konings, K. T. S., & Stoffers, H. E. J. H. (2018). Competence of general practitioners in requesting and interpreting ECGs-a case vignette study. Netherlands Heart Journal, 26(7), 377-384. https://doi.org/10.1007/s12471-018-1124-2 [4] Dadlani, G. H., Edwards, T. C., Fishberger, S., Epelman, A., & Erbrich, N. (2018). Pediatric electrocardiograms for the general practitioner: The importance of the T-wave. Pediatric Annals, 47(3), 106-111. doi:http://dx.doi.org/10.3928/19382359-20180219-01 [5] Ashley EA, Niebauer J. Cardiology Explained. London: Remedica; 2004. Chapter 4, Understanding the echocardiogram. Available from: https://www.ncbi.nlm.nih.gov/books/NBK2215/

Built With

• julia