Coronavirus disease 2019-associated encephalitis and concomitant subdural hematoma: a case report.

Bipolar Disorder (BD) is characterised by various medical conditions related to mental health, which sometimes are quite challenging to differentiate and distinguish from other mental disorders. The majority of existing screening approaches of BD are based on questionnaire-based tools as well as an interview process that results in either image, video, or text-related information, and sometimes it also includes radiological data too. All the screening methods are based much time consuming that demands multiple visits to the doctor. However, early detection of BD is necessary for enhanced treatment outcomes, prevention of disability, minimised threat of comorbidities, better prognosis, etc. Machine Learning (ML) has recently contributed towards addressing multiple ranges of problems associated with mental health disorders. However, there is still an unsolved challenge too. Data associated with BD could be very scarce, especially in early phases of BD, while it can also be inconsistent, too, which is definitely not enough to build any training model that eventually lacks diversity. There is often a cyclical pattern of mood for an individual with BD, which could be quite complex, too, posing enough challenge for a standardised model to carry out forecasting. Finding the progression point of BD is another bigger challenge. From a practical viewpoint, the prominent impediment also resides in integration ML based approach with clinical settings. Adoption of Artificial Intelligence (AI) in sensitive medical information is something that many healthcare professionals may be reluctant to use for diagnosis, especially when there is no ground-breaking, proven approach to date. Hence, the proposed study presents a novel and cost-effective solution to address the ongoing challenges associated with early detection of BD. This paper presents a novel multimodal feature promoting and enriching the profiling of the normal and BD participants for better identification of indicators using unique intelligent modelling. The proposed computational model depicts a risk screening tool that is essentially meant for leveraging early monitoring rather than any form of definitive system for clinical diagnosis. Although behavioral screening performed by an AI, it is essential to realize that proposed model depicts an explanatory model assessed on cohort that is relatively of limited size performed on highly controlled research environment. Hence, the accomplished outcome should be understood as universal generalizability or an evidence of clinical validation. The key motive of IDMBD is towards analyzing and exploring the possibility towards early screening of BD adopting multimodal behavioral analytics. Yet, the study considers validating with large-scale clinical environment as next phase of future work direction. The novelty of this paper is that it presents a cost-effective framework that doesn’t demand exclusive resources or platforms or any extra cost towards deployment or maintenance, harnessing a hybrid form of an innovative learning strategy different from existing solutions in the present times.