A method for rapid machine learning development for data mining with doctor-in-the-loop.

Classifying free-text from historical databases into research-compatible formats is a barrier for clinicians undertaking audit and research projects. The aim of this study was to (a) develop interactive active machine-learning model training methodology using readily available software that was (b) easily adaptable to a wide range of natural language databases and allowed customised researcher-defined categories, and then (c) evaluate the accuracy and speed of this model for classifying free text from two unique and unrelated clinical notes into coded data. A user interface for medical experts to train and evaluate the algorithm was created. Data requiring coding in the form of two independent databases of free-text clinical notes, each of unique natural language structure. Medical experts defined categories relevant to research projects and performed 'label-train-evaluate' loops on the training data set. A separate dataset was used for validation, with the medical experts blinded to the label given by the algorithm. The first dataset was 32,034 death certificate records from Northern Territory Births Deaths and Marriages, which were coded into 3 categories: haemorrhagic stroke, ischaemic stroke or no stroke. The second dataset was 12,039 recorded episodes of aeromedical retrieval from two prehospital and retrieval services in Northern Territory, Australia, which were coded into 5 categories: medical, surgical, trauma, obstetric or psychiatric. For the first dataset, macro-accuracy of the algorithm was 94.7%. For the second dataset, macro-accuracy was 92.4%. The time taken to develop and train the algorithm was 124 minutes for the death certificate coding, and 144 minutes for the aeromedical retrieval coding. This machine-learning training method was able to classify free-text clinical notes quickly and accurately from two different health datasets into categories of relevance to clinicians undertaking health service research.

Effect of Core Temperature and Embolic Load During Cardiac Surgery on Motion Perception.

BACKGROUND: Cognitive decline post-cardiac surgery is of clinical concern. To better understand it a sensitive and specific measure of post-surgery brain impairment is required. The cerebral territory most likely to be adversely affected by surgery is the posterior "watershed" territory. METHODS: We have designed a psychophysical task involving reading letters defined by motion aimed at measuring the integrity of a cortical area (MT) located in this territory. Patients undergoing coronary artery bypass grafting (CABG) and a healthy control group were given the psychophysical test twice, pre- and post-surgery for the patient group. RESULTS: There was no overall difference in performance between the surgery group and the control group at either pre- or post-surgery testing. However, multivariate analysis of surgical variables such as body temperature and embolic load to the brain as measured by Transcranial Doppler showed that patients with warmer core temperatures and higher embolic loads performed significantly worse on the motion defined letter reading tasks than those with more favourable surgical variables. CONCLUSIONS: These results demonstrate that high embolic load and warm core body temperatures lead to poor motion perception post-cardiac surgery, implying damage to the posterior watershed cortex.

Cue gradient and cue density interact in the detection and recognition of objects defined by motion, contrast, or texture.

The human visual system is able to extract an object from its surrounding using a number of cues. These include foreground/background gradients in disparity, motion, texture, colour, and luminance. We have investigated normal subjects' ability to detect objects defined by either motion, texture, or luminance gradients. The effects of manipulating cue density and cue foreground/background gradient on both detection and recognition accuracy were also investigated. The results demonstrate a simple additive relationship between cue density and cue gradient across forms defined by motion, luminance, and texture. The results are interpreted as evidence for the notion that form parsing is achieved via a similar algorithm across anatomically distinct processing streams.