A New Tool from the State Data Agency: A Solution for Cholesterol Test Analysis
There was a time when the joke went: “How do you recognize a doctor? By an illegible handwriting”. Today, some of these “hard-to-read” medical records (metaphorically speaking) persist, but in an electronic system. The State Data Agency has taken on the challenge of “deciphering” part of this data, specifically related to cholesterol tests for the population. The Agency’s specialists have developed a new tool that provides access to highly accurate cholesterol test data for the country’s residents. This innovation is expected to enable researchers to improve treatment methods and reduce the prevalence of cardiovascular diseases.
“Recently, the State Data Agency has undertaken several projects related to the spread of cardiovascular diseases and cholesterol testing in our country. When there is a need from the scientific or business community, we always assist in selecting and preparing data tailored to specific objectives. Researchers from the Lithuanian University of Health Sciences and its hospital, Kaunas Clinics, reached out regarding these specific data needs. The health data team at the Agency, as part of these projects, developed a tool designed to analyze various records stored in the E-health system and systematize laboratory results of so-called “bad” cholesterol tests,” explains Dr Julius Juodakis, Head of the Agency’s health data team, about the new project.
An Algorithm for Free Text Analysis Has Been Developed
Currently, after conducting laboratory tests, medical professionals input the results into the eHealth system in free-text form: copying, rewriting, and adding descriptions of symptoms and treatment recommendations. But how to identify those crucial data amidst a multitude of entries, particularly those vital in combating cardiovascular diseases, the leading cause of mortality?
Specialists from the State Data Agency have developed a set of algorithms designed to pinpoint cholesterol laboratory test results within diverse medical records. These results are identified using specific keywords, dates, and refined terms in both the Lithuanian and English languages.
According to the Agency’s specialists, this newly developed tool identifies “bad” cholesterol values with approximately 95 per cent accuracy. The algorithm can recognize various recording methods, such as when one doctor separates date digits with dashes, while another uses dots. The remaining few per cent include results recorded in unclear or highly unusual ways, such as when more than one value is written next to the test name.
“Why didn’t we use artificial intelligence for this task? Our answer is simple: medicine is a sensitive field where trust and transparency are paramount. We have developed our set of algorithmic rules to achieve the highest possible accuracy. Our team processed 230 million free-text records and identified 1.7 million entries on cholesterol tests, complete with measurements and test dates”, that is the explanation of Simona Zubavičiūtė, the Agency’s analyst and algorithm’s author.
Useful for Other Research
Specialists at the State Data Agency can now not only provide “bad” cholesterol values, but also integrate this data with other state records to address a variety of tasks. For example, researchers can investigate how changes in patients’ test results are influenced by the use of statins, i.e. medications designed to regulate “bad” cholesterol levels. It is also possible to link test results with patient diagnoses, mortality rates, and other data. Pharmacists might find it relevant to determine whether patients are taking the prescribed medications and whether they prefer more expensive or cheaper alternatives.
“The use of cholesterol test data can vary widely depending on researchers’ goals. Importantly, this tool can also be utilized to extract other test indicators from the eHealth system, further expanding its application possibilities,” notes S. Zubavičiūtė.
A Tool Used for Academic Research
Viktorija Butrimaitė, PhD candidate at the Lithuanian University of Health Sciences, in her doctoral thesis on pharmaceutical care, focuses on the pharmacist’s role in educating a patient about cardiovascular diseases, the prevalence of “bad” cholesterol, and its causes: one of the most common being the failure to take prescribed medications.
“My vision is for doctors and pharmacists to work as a team. A doctor prescribes the medication, while a pharmacist, upon dispensing it, could provide additional advice and explain how the medication might impact a chronic condition. The main issue is that patients tend to listen to comments from others about potential side effects and ultimately decide not to take what has been prescribed. For people, medication safety is especially important, and a pharmacist could be the best advisor, offering patients the necessary information,” explains V. Butrimaitė.
Researching Statin Usage
PhD candidate V. Butrimaitė is focusing her research on the use of statins, medications designed to lower “bad” cholesterol levels, due to a worrying tendency: cardiovascular diseases remain the leading cause of death, and their prevalence is not declining. Moreover, statin usage has not yet been measured on a national scale in Lithuania, making it impossible to compare our country’s data with that of other EU nations.
“Although statins are the best preventive measure for addressing cholesterol-related issues, people often do not take them. If a medication is prescribed, the pharmacist can see it in the E-health system. However, if the patient visiting the pharmacy does not intend to purchase the medication, the pharmacist could explain why taking it is important – this is the goal I aim for. It’s worth noting that about 50 per cent of people worldwide do not take the medications prescribed to them,” states V. Butrimaitė.
The tool developed by the State Data Agency, which enables nationwide access to cholesterol laboratory test results, provides PhD candidate V. Butrimaitė with a solid foundation to substantiate her hypotheses, predict trends, and draw data-driven conclusions.
“The statistics at my disposal allow me to see a very realistic picture. I hope that by completing my doctoral thesis, I will also discover a model for how pharmacists can best contribute to addressing this situation,” states V. Butrimaitė.
Last updated: 25-11-2024
Related news:
We Invite You to Submit Proposals for the 2027 Official Statistics Programme
OECD and the EU Invite Participation in the International R&D Careers Survey
Another State Data Lake Training Cycle Completed
How to Recognize a Survey Interviewer from the State Data Agency (Statistics Lithuania)?