Analytics Theories for Medical Diagnosis

Khivsara (2018) presents a number of basic analytics theories. Of these, I believe four are most relevant for medical diagnosis: clustering, association rules, regression, and textual analysis.

No alt text provided for this image

Association rules are nothing more than finding casual structures and patterns between objects in order to establish some sort of logical relationship. It is a machine learning analog to what doctors do on a regular basis in making diagnoses. Picture an emergency room triage room, where patients are sorted and prioritized based on symptoms. In place of a nurse, perhaps on particularly busy nights, a self-service kiosk would allow patients to select all the symptoms they are exhibiting and these symptoms would generate potential diagnoses, the severity of which would determine priority in the night’s order.

Moving a step beyond simple associations, let us examine clustering. Assume two risk factors for chronic disease (e.g., unhealthy diet and tobacco use) were quantified for a population of patients and plotted on a two-axis graph. A simple review of the graph would show plots of individuals on the spectra of diet and tobacco use. Rather than being evenly dispersed across the graph, the data points would be arranged in two or more groupings depending upon the population. K-means clustering would classify those data points (the individuals) into different risk groups depending upon where they fell on the chart. K-means clustering is most useful in healthcare applications where similarities between patients must be quantified and cohorts established.

Going a step further and putting quantitative measures on the relationship between variables and predicted values, we have regression. Regression is all about quantifying the relationship between sets of variables and predicting values. In healthcare, the most common use of regression is related to healthcare costs. Insofar as making diagnoses, logistic regression in particular can be helpful with making diagnoses based on a number of known factors. Imagine a known regression equation for predicting diabetes risk based on multiple input variables.

Finally, let us examine textual analysis. The other three theories mentioned here rely on structured data. However, that structured data is only a fraction of the data collected when a patient sees a provider. The ability to utilize the unstructured data, rife with context and nuance, is perhaps the biggest untapped potential in healthcare analytics. The confluence of textual analysis and natural language processing (NLP) allow unstructured data from sources such as patient records and provider dictation to become part of the picture in predictive modeling and coexist with structured data.

References

EMC Services. (2018). Data Science and Big Data Analytics: Discovering, Analyzing, Visualizing and Presenting Data. Retrieved from https://bhavanakhivsara.files.wordpress.com/2018/06/data-science-and-big-data-analy-nieizv_book.pdf

Healthcare.AI. (2017). Step by step to K-Means clustering. Retrieved from https://healthcare.ai/step-step-k-means-clustering/

HealthCatalyst. (2019). How to use text analytics in healthcare to improve outcomes. Retrieved from https://www.healthcatalyst.com/how-to-use-text-analytics-in-healthcare-to-improve-outcomes

Kulkarni, A. R., & Mundhe, S. D. (2017). Data mining technique: An implementation of association rule mining in healthcare. International Advanced Research Journal in Science, Engineering and Technology, 4(7), 62-65.

World Health Organization. (2005). Chronic diseases and their common risk factors. Retrieved from https://www.who.int/chp/chronic_disease_report/media/Factsheet1.pdf

The Privacy Divide: Social Media and Personal Genomic Testing

No alt text provided for this image

With every advance in technology comes a trade-off of some kind. Where the use of personally-identifiable information is concerned, the trade-offs typically involve the exchange of privacy and confidentiality for a non-monetary benefit. In the early days of social media, conventional wisdom said the product was the service. However, we have seen over the last decade that the users of such platforms are the products, the perceived benefits merely carrots on sticks to keep the products (users) engaged in the cycle. We willfully pour details of ourselves into various social media outlets, despite the documented bad behaviors by giants like Facebook, and mostly remain complacent in having our personal data packaged and leveraged against us by various business interests.

However, in the conversation I’ve had around personal genomic testing (PGT), I’ve noticed that many are quick to cite data privacy and risk as a key reason not to participate. Think about this. On one hand, we have evidence to prove Facebook has been using our data in dubious ways, yet we keep pouring ourselves into it (McNamee, 2019). On the other hand, the potential benefits of PGT are outweighed by a fear of that data potentially being misused.

My purpose is not to minimize the potential hazards around PGT. Consider the following risks: (a) hacking; (b) profit or misuse by the company or partners; (c) limited protection from a narrow scope of laws; (d) requests from state and federal authorities; and (e) changing privacy policies or company use due to mergers, acquisitions, bankruptcies, et cetera (Rosenbaum, 2018). In the face of potential benefits from PGT, these are serious caveats. But read that list outside of this context, and it is equally applicable to the data we generate and provide to social media outlets on a daily basis.

As of yet the privacy regulations around social media use only exist within the context of the company itself—that is, there are no substantial federal regulations in the US on the matter, only the GDPR in the EU (St. Vincent, 2018). Where health information is concerned, the US does have slightly more mature federal regulation. The Health Insurance Portability and Accountability Act (HIPAA) requires confidentiality in all individually-identifiable health information; in 2013, this law was extended to genetic information by way of the Genetic Information Nondiscrimination Act (GINA). While the rules prohibit use of genetic information for underwriting purposes, there is no restriction on the sharing or use of genetic information that has been de-identified (National Human Genome Research Institute, 2015). De-identification is not entirely foolproof. There are cases in which the data can be re-identified (Rosenbaum, 2018).

The incongruence is puzzling. In the case of social media, users willfully provide a wealth of data points on a regular basis to companies that repackage and monetize that data for dubious purposes, in the absence of meaningful US legislation to protect it. In the case of PGT, where at least HIPAA and GINA have a rudimentary level of codified protection, users’ hesitance appears to be much more pronounced.

References

McNamee, R. (2019). Zucked: Waking up to the Facebook catastrophe. New York: Penguin.

National Human Genome Research Institute. (2015). Privacy in genomics. Retrieved from https://www.genome.gov/about-genomics/policy-issues/Privacy

Rosenbaum, E. (2018). Five biggest risks of sharing your DNA with consumer genetic-testing companies. Retrieved from https://www.cnbc.com/2018/06/16/5-biggest-risks-of-sharing-dna-with-consumer-genetic-testing-companies.html

St. Vincent, S. (2018). US should create laws to protect social media users’ data. Retrieved from https://www.hrw.org/news/2018/04/05/us-should-create-laws-protect-social-media-users-data

Where Clinical, Genomic, and Big Data Collide

One of the early proving grounds of big data is healthcare, and the constant cycle of insights catching up to volume hasn’t changed since the early days of the electronic patient record. Early healthcare data typically involved structured metrics such as ICD9 codes and other billing data, which yielded very little clinical detail. The introduction of new data points, both structured and unstructured, has opened the door to many new analytics possibilities. While the possibilities are there, “few viable automated processes” exist that can “extract meaning from data that is diverse, complex, and often unstructured” (Barlow, 2014, p. 18). Indeed, the gap continues to widen between the “rapid technological process in data acquisition and the comparatively slow functional characterization of biomedical information (Cirillo & Valencia, 2019, p. 161).

With so much available, a hospital or healthcare provider may find it difficult to determine a place to start, and either ignore the possibilities altogether or engage in initiatives that are not impactful to clinical quality or costs. There are five broad areas in which value can be delivered: clinical operations, payment & pricing, R&D, new business models, and public health; data are gathered from four broad sources including clinical, pharmaceutical, administrative, and consumer (Barlow, 2014, p. 21).

As of late, genomics have entered the conversation as both a consumer product (e.g., 23AndMe or Ancestry, known as personal genomic testing) and clinical practice. It is one thing to prescribe a medication based on a patient’s chart history, but an entirely different patient experience when a prescription is tailored to a patient’s particular metabolism, genetic predispositions, and risks (Barlow, 2014, p. 19). The wealth of patient-generated health data from a growing number of consumer devices has already contributed to the rise of “Personalized Medicine” (Cirillo & Valencia, 2019, p. 162) and the introduction of genomic data will move the needle even further. One can’t get much more personalized than a genetic footprint.

One debate around personal genomic testing is the value it provides when given directly to consumers without the benefit of clinician involvement. While the benefits of such testing include lifestyle changes that mitigate future disease risk, consumers are also prone to misinterpretation that may lead to unnecessary medical treatment (Meisel et al., 2015, p. 1). Beyond future risk, a recent study found the interest around personal genomic testing had a great deal to do with family or individual history of a particular affliction (Meisel et al., 2015). Consumers are mindful of explaining current risks and phenomena, not just predicting them.

References

Barlow, R. D. (2014). Great expectations for big data. Health Management Technology, 35(3), 18-21.

Cirillo, D., & Valencia, A. (2019). Big data analytics for personalized medicine. Current Opinion in Biotechnology, 58, 161-167.

Meisel, S. F., Carere, D. A., Wardle, J., Kalia, S. S., Moreno, T. A., Mountain, J. L., . . . Green, R. C. (2015). Explaining, not just predicting, drives interest in personal genomics. Genome Medicine, 7(1), 74.