If existing algorithms are not sufficient for your research question(s), I can develop a new algorithm for you. I have extensive experience with the development of algorithms for accelerometer data, including: signal error reduction, missing data imputation, energy expenditure estimation, activity type detection, and sleep detection. I am familiar with both physiological and biomechanical concepts as well as the fundamentals of machine learning and statistics. A full algorithm development trajectory typically involves the following steps: Literature research, Data collection, Data quality check, Exploration of variety of classification techniques (incl. machine learning), Proof of concept, Cross validation and Publication. I can provide services around all of these stages.
I am able to build new Open Source research software as well as enhance existing software. This can include data management tasks, algorithm implementation, data visualisation functionalities, as well as setting up a good software development infrastructure. I have experience with both R and Python, work both in Windows and Linux, and share my code with version control.
Dr. van Hees holds a PhD from University of Cambridge (UK), and developed and published the now widely used methods for automatically correcting for signal calibration error, signal component separation for accurate physical activity assessment, and methods for sleep detection. Further, he worked as Senior Research Engineer at the Netherlands eScience Center (Amsterdam). Throughout the years he has worked with a broad variety of technologies, including: R, Python, git, SQL, machine learning (incl. deep learning), Inertial measurement units, Polysomnography, and Indirect calorimetry.