Innovations in Synthetic Electronic Health Records: Opportunities and Challenges

Challenges Offered by Synthetic EHRs

Enhanced Privacy and Security

A major plus to synthetic EHRs is the improved privacy and security that comes with synthetic EHRs. Another advantage of synthesizing EHRs is that there is no possibility of repeating the genuine reader’s privacy violation, as synthetic EHRs do not consist of patients’ real information. This in turn makes it easier for the researchers and healthcare providers to share or analyze data without the herculean task of de-identifying patients’ information. The utility of synthetic EHRs also eliminates the legal and ethical issues of managing patient information, which opens the possibility of conducting research under the regulation of HIPAA, for instance.

Increased Accessibility for Researchers

Realistic synthetic EHRs offer to research the kind of big and diverse data sets that are needed for a variety of purposes, as such datasets can be obtained from electronic health records. Historical EHRs are sometimes locked in silos imposed by security mechanisms and/or patients’ consent, which may hinder the availability of data to researchers. Structurally, synthetic EHRs differ from real EHRs, and while the latter can be shared and used in research without patients’ consent, the former can best be described as invaluable for researchers. This ease can in turn mean that the advancement of medical research could also be faster due to the overall quickening of emerging health knowledge in the world.

Higher Data Quality and Some Measures of Standardization

As opposed to real patient data, synthetic EHRs can be tailored to fit the needs of the given study and guarantee high quality, as well as the uniformity of the data provided. This is especially true where data precision is of the essence, which implicates fields such as clinical trials. With synthetic EHRs, some of the real-world noises can be avoided, hence giving more accurate and consistent results to the researchers.

Facilitating the Development of Predictive Models

The lifelikeness of such EHR can make a highly positive impact when it comes to the creation and improvement of predictive models in the overall framework of healthcare. These models base their information analysis on historical data to predict the probable outcome of the patient, a process known as data mining. Synthetic EHRs are an ideal dataset, especially for the development of the models as they can be simulated a priori in large numbers and addressed to conditions or populations of interest. This can make the generation of better prediction models possible, which in the long run enhances the treatment type offered to patients.

Facilitating In-Silico Clinical Trials

Mathematical-based models and real-life patient virtual models used in what is referred to as in-silico clinical trials are gunning to be the new frontier in drug development. Among them, synthetic EHRs are significant because they generate data that is used to build virtual patients for the trials. This enables the investigators to introduce new therapies in a suitable and less risky environment than on humans or animals. The in silico trials can also be done at a faster pace and actual cost than the clinical trials, which makes it even more appealing to the pharmaceuticals.

Challenges Facing Synthetic EHRs

Nevertheless, there are some concerns arising from the synthetic EHRs, which need to be addressed. such as:

Ensuring Data Realism and Validity

A key concern with synthetic EHRs is that of external validity, that is, how realistic the generated EHRs are. The main drawback of using synthetic trials is that they are built up to mimic the real EHR data, and in some cases, they can be disparate from the real environment. This can be especially so in situations, most of which have so many components that their relations become hard to foresee. The need to keep the synthetic EHRs more realistic and valid means that more studies and validation have to be conducted to achieve high data realism and validity.

Limitation of Technical and Computational Facilities

The construction of artificial EHRs is based on the use of the algorithm and computational model, which entails significant costs and technical complexities. Building these models needs a proper knowledge of healthcare and statistical modeling, and often a slight discrepancy in the decision process leads to a massive difference in the results. Also, creating large synthetic data sets takes a lot of computation power, and hence using synthetic EHRs can be off-putting in terms of costs, especially for small institutions.

Addressing Ethical Considerations

On the one hand, using synthetic EHRs benefits from the privacy and security perspective; on the other hand, using fake data in research raises ethical concerns. For instance, there is a chance that scientists may use synthetic data as presidents do political cards; only reliance on synthetic data will lead to conclusions that are irrelevant to regular patients. Further, the use of synthetic EHRs in clinical trials makes it seem like the validity of those trials is compromised along with the impacts in terms of patient safety that can result from it.

Integration with Real World Data

One of the issues, thus, related to the employment of synthetic EHRs is the possibility of combining them with real-world data. As can be seen, there are several advantages to using synthetic EHRs, even though they are not an ideal replica of the actual data. There are challenges in merging such synthetic EHRs with real-world data since there may be disparities between the two types of data, and therefore the combination of real-world data with synthetic data will entail resolving such disparities and ensuring that the results are genuine. This is especially relevant for times when synthetic EHRs introduced as an approximation to the real ones are utilized instead of or in addition to the genuine patient information.

Regulatory and Standardization Issues

Synthetic EHRs are far from being perfect mimics, and the use of synthetic EHRs in healthcare and research is itself a new discipline, and therefore there are hardly any guidelines governing their usage. This may result in disparities in the quality and reliability of synthetic EHRs, which is not ideal for use when determining the accuracy of data used by researchers and healthcare providers. The creation of regulatory control measures and standardization of best practices for synthetic EHRs will be of utmost importance for their safe practice in healthcare systems.

Conclusion

Electronic health records are one type of synthetic data that was introduced in the healthcare industry as a very innovative tool, which has many possibilities for developing research, clinical trials, and patient care. Nonetheless, achieving the optimum utility of synthetic EHRs will be possible only if other monumental issues related to data realism, technical constraints, and ethical and regulatory compliance are resolved. Further advances in synthetic EHRs will make it important for those who specialize in this area, the medical practitioners, and government regulators to collaborate and bring out all such possibilities that will ensure the use of this particular technology in health care is beneficial rather than having adverse outcomes. With due dedication, synthetic EHRs hold the potential to become the pillars of today’s healthcare framework: effective research, superior patient care, and far more secure and efficient systems.

References

1. Walonoski, J., Hall, D., Bates, K.M., Farris, M.H., Dagher, J., Downs, M.E., Sivek, R.T., Wellner, B., Gregorowicz, A., Hadley, M. and Campion, F.X., 2022. The “Coherent Data Set”: Combining patient data and imaging in a comprehensive, synthetic health record. Electronics, 11(8), p.1199.
2. Walonoski, J., Kramer, M., Nichols, J., Quina, A., Moesel, C., Hall, D., Duffett, C., Dube, K., Gallagher, T. and McLachlan, S., 2018. Synthea: An approach, method, and software mechanism for generating synthetic patients and the synthetic electronic health care record. Journal of the American Medical Informatics Association, 25(3), pp.230-238.
3. McLachlan, S., Dube, K., Gallagher, T., Daley, B. and Walonoski, J., 2018. The ATEN framework for creating the realistic synthetic electronic health record.
4. Pappalardo, F., Russo, G., Tshinanu, F.M. and Viceconti, M., 2019. In silico clinical trials: concepts and early adoptions. Briefings in bioinformatics, 20(5), pp.1699-1708.
5. Na, L., Yang, C., Lo, C.C., Zhao, F., Fukuoka, Y. and Aswani, A., 2018. Feasibility of reidentifying individuals in large national physical activity data sets from which protected health information has been removed with use of machine learning. JAMA network open, 1(8), pp.e186040-e186040.
6. Benaim, A.R., Almog, R., Gorelik, Y., Hochberg, I., Nassar, L., Mashiach, T., Khamaisi, M., Lurie, Y., Azzam, Z.S., Khoury, J. and Kurnik, D., 2020. Analyzing medical research results based on synthetic data and their relation to real data results: systematic comparison from five observational studies. JMIR medical informatics, 8(2), p.e16492.
7. Texier, G., Jackson, M.L., Siwe, L., Meynard, J.B., Deparis, X. and Chaudet, H., 2017. Building test data from real outbreaks for evaluating detection algorithms. PloS one, 12(9), p.e0183992.
8. Sun, Z., Wang, F. and Hu, J., 2015, August. LINKAGE: An approach for comprehensive risk prediction for care management. In Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (pp. 1145-1154).