Building an Empirical Treatment Protocol from High-Resolution Traumatic Brain Injury Data

Stell, A., Moss, L. and Piper, I.

An informatics issue common for many fields of medical research is the poor standardisation of baseline clinical management data, which can have a large negative impact on the statistical power of drug studies making use of that data. This baseline variation can be for many reasons e.g. rarity of the condition but, despite the development of standardised medical guidelines in many areas, it is still often the case that study data is affected by the \"real world\" differences in treatment protocols. To improve understanding of that management baseline in general, this paper describes work that builds up an empirical treatment pattern from retrospective intensive care unit (ICU) data. The ultimate goal is to build protocol 'objects' that can be compared between specialist centres or 'gold standard' guidelines. Variation and differences between these objects can then be quantified and potentially mitigated - to allow a more standardised comparison of data for studies, as well as providing information on audit and guideline adherence. From a combination of event detection from high-resolution physiological output and association of those detected events with annotated treatment information, an empirical data-driven notion of treatment protocols across specialist centres can be built. Using data drawn from Traumatic Brain Injury (TBI) studies, the initial steps of this technological work including the algorithms and assumptions of these two key functions are presented. The results when applied to a specific TBI data-set (Piper et al 2010) show how the event numbers vary when key parameters are changed (e.g. the hold-down time) and how this impacts clinical decisions and trial conduct.

Cite as: Stell, A., Moss, L. and Piper, I. (2014). Building an Empirical Treatment Protocol from High-Resolution Traumatic Brain Injury Data. In Proc. Seventh Australasian Workshop on Health Informatics and Knowledge Management (HIKM 2014) Auckland, New Zealand. CRPIT, 153. Warren, J. and Gray, K. Eds., ACS. 79-88

(from crpit.com) (local if available)