Answering the dosage question through groundbreaking data science
A renowned authority in healthcare manufacturing, the client wanted to improve dosing for an inhaler-delivered medicine. As a globally popular product, there was significant variation among the population using it. An incorrect dosage could result in medicine was ineffective or worse, dangerous. The key was to find a safe medium that adequately served all users – a task Centific was engaged to accomplish.
Applying data science to a large body of biosensor data, the Centific team was able to collect and correlate a range of critical user metrics, which were then used to create a safe and effective dosage plan. By the time of project close, the client would have a body of useful information to inform its medicine production, while physicians and patients alike would enjoy more accurate medication usage instructions informed by large sets of relevant data.
KEY SUCCESSES
- Analyzed biosensor data to develop an effective dosage model
- Employed machine learning systems for efficiency and accuracy 30% boost in training completions
- Developed an intelligent, predictive system around individual dosage
PUTTING BIOSENSOR DATA TO WORK
When the client engaged Centific, it had been collecting data through the use of biosensor technology. Of this data, the metrics most important to the team’s goals were blood oxygen level (SpO2) and heart rate, though a wide range of other biometrics were also considered during research.
Applying industry-leading knowledge in data science, the Centific team analyzed and activated this data to explore the relationship between patient behavior and dosage amounts.
This approach offered a distinct benefit: By monitoring biometrics throughout trials, our team could make informed decisions regarding what stats were important to the dosage question – and which weren’t.
MACHINE LEARNING SOLUTIONS
With the numbers crunched, our team was able to turn data insights into powerful, machine learning-driven tools for prediction and decision-making.
Notable among these was a Generalized Regression Neural Network (GRNN), developed by our team of emerging technology experts, which provides intelligent resources for several business processes, including prediction and classification. Similarly, a Bootstrap Aggregating Decision Tree – also known as a “bagged” decision tree – was developed to bolster the stability and accuracy of the machine learning infrastructure employed by the Centific team.
Together, these tools could predict inhaler volume of a single user based on biosensor data, relying on big data and iterative trials to hone the accuracy of their results.
RESULTS
Medicine efficacy relies on proper dosing. Depending on our team wasn’t just client success, but also the wellbeing of the patients that use the product and the physicians that prescribe it. By leveraging our experience in data science, Centific acted as a trusted partner for a major healthcare client. The result: more accurate dosage guidelines and happier patients.