There are many challenges with healthcare data that make it difficult to capture and measure. LeSueur (2014) pointed out five: 1) data is in multiple places, 2) structured and unstructured data, 3) inconsistent variables/definitions, 4) data is complex, and 5) changing regulatory requirements. There are other challenges such as dealing with sensors, body changes, and standardization problems. There are ways to iron out problems of healthcare data.
Data that derives from sensors with the Internet of Things (IoT) can be problematic. These sensors are worn constantly and are measuring a constantly changing human body. Although computing technology is much cheaper than before, the variety of data is still staggering. There could be XML from one data source but an image from another (Firouzi et al., 2018). One proposed solution is to have a cloud/fog administrator that specializes in analyzing these different databases. Another has been to have the device sensors integrated with the computing and networking. The latter seems more practical as this allows the data to be more readily available and not transferred to a cloud system that may be overburdened.
The other problem is the lack of standardization. There are many initiatives trying to create standardizations such as openEHR, SNOMED CT, LOINC,HL7, FIHR, and OBO (Mansmann et al., 2017) . However, due to the nature of the data, standardization will take time. About 75% of our data is unstructured and resides in text files (Mastrian, 2017, p. 312). Furthermore, many data-mining algorithm don’t work well with text (Jamsa, 2021, p. 529). The best solution to text data would be the process of finding keywords and patterns much to what a Google search engine would do. Text data requires a meticulous cleaning process as well as more specificity in the question the analyst is trying to ask.
LeSueur (2014) is correct in assessing that data will be more complex but there will be more tools in place that will make things easier to use. Much like the evolution from traditional statement based programing to a more GUI based operating system, our computers will be better prepared to deal with the complexity. We will be better prepared for the data over time.
References
Firouzi, F., Farahani, B., Ibrahim, M., & Chakrabarty, K. (2018). Keynote Paper: From EDA to IoT eHealth: Promises, Challenges, and Solutions. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 37(12), 2965-2978.
Jamsa, K. (2021). Introduction to data mining and analytics : with machine learning in R and Python. Burlington, MA: Jones et Bartlett learning.
LeSueur, D. (2014, June 12). 5 Reasons Healthcare Data Is Unique and Difficult to Measure. Retrieved from https://www.healthcatalyst.com/insights/5-reasons-healthcare-data-is-difficult-to-measure
Mansmann, U. (2017). C. D.COMBS, JOHN A.SOKOLOWSKI and CATHERINE M.BANKS. The Digital Patient: Advancing Healthcare, Research, and Education. Hoboken: Wiley. CHANDAN K. REDDY and CHARU C. AGGARWAL. Healthcare Data Analytics. Boca Raton: CRC Press. Biometrics, 73(4), 1467-1468.
Mastrian, K. G. (2017). Informatics for health professionals. Burlington, MA: Jones & Bartlett Learning.