Sleep scoring software written during Phase I using robust waveform identification techniques will be further developed to support graphical """"""""adaptive analyses"""""""". Existing stage scoring programs do not allow analysts to alter sleep stage decision criteria. Some permit changing an epoch's score, but re-analysis of the same data montage merely regenerates the original, incorrect scores. All FDA-approved scoring software perform acceptably on their validation data sets, but problems inevitably have arisen when confronted with the great variety of real- world data. Analysts rapidly realize these shortcomings and the programs fall into disuse. Further, existing programs require extensive re-coding when new sleep parameters are introduced. We propose creation of truly adaptive analysis software for clinical and research markets: allowing graphic access to decision criteria, configurations may be easily tailored and saved which reflect different disorders, age dependencies, or individual analyst preferences. Even criteria involving completely new data types will be supported. By these means we will overcome basic limitations of existing sleep analysis programs: their inherent technical limitations and analytic inflexibility. At the end of Phase II we will have developed and tested in respected sleep research labs a commercial-quality sleep study program able to meet present and future analysis needs.
Beyond the intended market in sleep medicine, the """"""""adaptive analysis"""""""" capability has commercial potential for other types of research and clinical environments, animal sleep studies, man-in-the-loop industrial process control, or any field requiring user-tailored detection and analysis of multi-channel events.
