Design and evaluation of a mobile decision support system for screening neurodevelopmental disorders
|dc.contributor.author||Cullen, Brittany Rose|
|dc.identifier.citation||Cullen, B. R. (2020). Design and evaluation of a mobile decision support system for screening neurodevelopmental disorders (Thesis, Doctor of Philosophy). University of Otago. Retrieved from http://hdl.handle.net/10523/10231||en|
|dc.description.abstract||New Zealand has the highest rate of youth suicide in the developed world. Young persons with ADHD and ASD, amongst other neurodevelopmental disorders, are at greater risk for depression, suicidal ideation, and suicide attempts than their typically developing peers. The early detection and treatment of neurodevelopmental disorders significantly improves the symptomatology and life outcomes of affected individuals. Gaining timely access to specialist mental health clinicians is difficult. This process is made more difficult for children who do not present with overt signs of severe and persistent psychological distress and therefore are not deemed to be ‘high risk’, including the majority of children with ADHD and ASD. Consequently, the diagnosis of children with neurodevelopmental disorders is substantially delayed and/or the responsibility to identify and/or treat falls on clinicians (and non-clinicians) such as general practitioners, general paediatricians, nurses, and educators who may not have the background to have developed complex schemas in this domain. The aim of the present thesis was to develop and test decision support systems that employ different reduced-processing design principles. This was done to determine which principles were effective in assisting individuals who do not possess complex clinical schemas, to diagnose children with neurodevelopmental disorders, in the absence of training and practice with the systems. Several manipulations of decision support system interfaces were developed and tested across different populations of prospective diagnosticians. Across four studies, participants were randomly assigned one of several diagnostic-aids, which differed in their capacity to reduce-processing demands, to assist with clinical diagnosis in a simulated context. Study One: The researcher tested the efficacy of the DSM-5 and two alternate diagnostic-aids that employed one or two reduced-processing design principles that pertained to the arrangement (facilitated the simultaneous acquisition of information) and/or volume (restricted content preconfigured by two clinicians who specialise in neurodevelopmental disorders) of diagnostic information in assisting naïve diagnosticians to diagnose two patients in two separate filmed clinical interviews (both 20-minutes in length). Participants who used the tablet-based diagnostic-aid that employed two (but not one) reduced-processing principles achieved significantly greater diagnostic accuracy scores, with the vast majority of participants in this group correctly (and more efficiently) diagnosing both a seven-year-old with ADHD and a 3-year-old with ASD. There were no significant differences in diagnostic accuracy or response latency between participants who used the DSM-5 and participants who used the diagnostic-aid that employed one reduced-processing principle (restricted, preconfigured content). Study Two: The paradigm from Study One was adopted in Study Two to test the efficacy of the DSM-5 and the diagnostic-aid that employed two reduced-processing design principles in assisting novice diagnosticians (Psychology Master’s students) to diagnose both ASD and ADHD. Participants who used the ‘reduced-processing’ diagnostic-aid achieved significantly greater diagnostic accuracy scores compared to participants who used the DSM-5, with the vast majority of participants in the ‘reduced-processing’ diagnostic-aid group making two correct diagnoses. Study Three: The paradigm from Studies One and Two was adopted in Study Three to examine the extent to which the interactive mechanism of the ‘reduced-processing’ diagnostic-aid to record categorised symptoms of perceived relevance to the patient, influenced diagnostic performance amongst naïve diagnosticians. Three versions of the ‘reduced-processing’ diagnostic-aid, which were identical in content volume and arrangement, were tested: An interface with an interactive mechanism to record categorised symptoms (the ‘Categorisation Interface’ that was used in Studies One and Two), an interface with an interactive mechanism to record uncategorised symptoms (‘Highlighting Interface’), and an interface with no recording mechanism (‘Passive Interface’). Participants who used the ‘Categorisation Interface’ achieved significantly greater diagnostic accuracy scores (and lower response latencies for correct diagnoses of ASD) compared to participants who used the DSM-5, ‘Highlighting Interface’, and ‘Passive Interface’, with the vast majority of participants in the ‘Categorisation Interface’ group making two correct diagnoses. There were no significant differences in diagnostic accuracy or response latency between the DSM-5, ‘Highlighting Interface’, and ‘Passive Interface’ groups. Study Four replicated and extended Study Three. The ASD scenario from the previous studies was adopted in Study Four to examine whether the ‘Categorisation Interface’ needed to be used ‘during + after’ the clinical footage (as it was in Studies 1-3), or if it could also be used ‘after’ the clinical footage to improve diagnostic accuracy. Participants who used the ‘Categorisation Interface’ immediately following the clinical footage were just as accurate (but significantly slower) in diagnosing ASD as participants who used the ‘Categorisation Interface’ both during and immediately following viewing the footage. Conclusion: A decision support system that is structured to facilitate the simultaneous acquisition of a restricted set of preconfigured critical diagnostic features, with the additional capacity to record categorised symptoms of perceived relevance to the patient, was found to assist individuals who do not possess highly developed (or complex) clinical schemas to accurately and efficiently diagnose both ASD and ADHD, in the absence of training and practice with the system. This research has important practical implications for the early detection of neurodevelopmental disorders.|
|dc.publisher||University of Otago|
|dc.rights||All items in OUR Archive are provided for private study and research purposes and are protected by copyright with all rights reserved unless otherwise indicated.|
|dc.title||Design and evaluation of a mobile decision support system for screening neurodevelopmental disorders|
|thesis.degree.name||Doctor of Philosophy|
|thesis.degree.grantor||University of Otago|
Files in this item
There are no files associated with this item.
This item is not available in full-text via OUR Archive.
If you are the author of this item, please contact us if you wish to discuss making the full text publicly available.