Apkar Salatian

A data wavelets approach to deriving trends in historical ICU monitor data

Medical staff in the Intensive Care Unit (ICU) are confronted with large volumes of continuous noisy data from several physiological sources that require interpretation. Rather than reasoning quantitatively on a point by point basis, especially in the context of other signals, we believe that Medical Staff could benefit with assistance in the interpretation of the ICU data by providing qualitative summaries. We propose data wavelets as an approach to analysing historical ICU data for deriving trends for summarization. In this paper we will show that wavelets are particularly effective for representing various aspects of non-stationary data such as trends, cycles and discontinuities

Data compression for ICU telemedicine to reduce transmission costs and allow qualitative reasoning

Intensive Care Unit (ICU) telemedicine can be considered as the transmission of large volumes of continuous and noisy physiological data generated by the ICU monitors attached to patients from one site to another using computer and telecommunication technology for purpose of remote assistance. A common form of telecommunication is broadband which presents 2 major challenges in rural areas: bandwidth demand can easily outstrip the revenue realizable that is needed to pay for the network infrastructure investment so lower (cheaper and slower) bandwidth is normal; a consequence of restricted bandwidth on access pipes is service contention at the customer site, even if core bandwidth exists to deliver the services. In this paper we propose data wavelets as a data compression technique to transform the ICU monitor data into trends to address the challenges of broadband in rural areas for data transmission and allow qualitative reasoning at the receiving site for clinical decision support.

An Expert System for Interpreting ICU Monitor Data

By performing analysis, we propose and describe an expert system called ASSOCIATE which analyses the historical voluminous high frequency and noisy data generated by the monitors of an Intensive Care Unit (ICU) for the purposes of summarisation and patient state assessment. ASSOCIATE is an associational reasoner and consists of 4 sub-systems: data filtering which is used to remove noise; interval identification to generate trends (intervals of time where the data is either increasing, decreasing or steady) from the filtered data; interpretation which performs summarisation and patient state assessment by applying rules from a knowledge base sub system to overlapping trends to identify clinically significant events.