The dissertation proposes a solution to automating the dosing titration process, which is the trial-and-error process involved in establishing a patient’s correct dose of medicine.
In his studies, Ilias employed quantitative methods and focused primarily on data mining methods and optimization algorithms. He designed two dosing algorithms, one for continuous infusion of medication and one for medication with tablets. He used data from wearable sensors and medical records from hospital admissions.
The dissertation comprises five papers: four have appeared in journals and one in conference proceedings. The papers describe how sensor data can be used to make individual dosing adjustments. Further, it examines how the effect of medication depends on the time of day.
Ilias’s first study describes how wearable sensors can be used to evaluate the medication status of the patients (undermedicated, properly medicated, and overmedicated). His second study uses the medication status information to evaluate patients’ dosing needs by creating individual patient profiles. In his second and third study, the two dosing algorithms are described and evaluated using the individual patient profiles. In the fourth study, the data from the first study are re-examined to improve the detection performance. The final study investigates how patients respond to a dose in the morning hours compared to the same dose in the afternoon hours. Results show that the response of patients to the same dose is lower in the afternoon.
Ilias’s dissertation demonstrates that the automation of the dosing titration process is possible. It is groundbreaking in that he is unaware of any other peer-reviewed study that proposes algorithmic dosing adjustments. Further, his study evaluates the feasibility of automating the dose titration process.
Ilias hopes that his dissertation will lead to further research that focuses on validating the system and refining the dosing algorithms and the design of other dosing algorithms.
Ilias states, “I think that this research is very important since its application will be necessary in future generations. With the aging population, the number of patients is expected to double in 20 years. That means that neurologists might not have enough time to treat patients and the dosing process needs to be automated so that time and money can be saved within the healthcare system.”
Most importantly, he believes that patients will benefit since they will be able to adjust doses at home, the result of which will be a better quality of life.
“The field of Microdata Analysis is about data-driven decision support. I feel that I really explored how this field applies to personalized medicine and that I might have discovered a future direction for Microdata Analysis studies,” he concludes.