Behavioral observation is an important tool for early diagnosis of many neurological disorders in children. Behaviors like repetitive movements, staring spells, and tantrums can be early signs of serious disorders like autism, epilepsy and ADHD. Such intermittent behaviors are difficult to capture during brief clinical visits, and accounts from caregivers are usually biased and imprecise. The holy grail of behavioral observation is continuous observation over days or even weeks in naturalistic environments such as home or school. Such a system is useful not only for diagnosis but also for tracking effectiveness of treatments and detailed functional analysis of problematic behaviors. Existing video behavior-coding systems cannot be used as they lack privacy protection of patients and bystanders as required by certain laws (the HIPPA and FERPA laws). This project focuses on the development and commercialization of an audiovisual recording and coding system for clinical use, specifically designed to capture human behaviors in naturalistic environments such as home and school.
This project focuses on the development and commercialization of the Video Interface for Behavioral Evaluation (VIBE) system. A key enabling feature of VIBE is its innovative use of visual privacy protection: a "privacy bubble" around the patient can be set up for observation, whereas the rest of the environment is blurred. All the stakeholders can jointly decide on the extent of the bubble to balance the privacy needs and the utility of the recorded video. In addition to the 'privacy bubble,' VIBE supports a wide array of automatic event filters that help clinicians to identify not only the behavioral episodes, but also the antecedent and the consequence of the behaviors, enabling the design of a more comprehensive treatment plan.
(VIBE) system. VIBE is an audiovisual behavior recording, coding, and summarization system for clinical use, specifically designed to capture human behaviors in naturalistic environments. Behavior observation is an important tool for a wide range of applications from security to medicine. Our focus is on the development an effective behavior observation system in naturalistic environments for research, diagnosis and treatment of neuro-developmental disorders like autism and ADHD in children. There is a significant market need due to an alarming increase of these disorders in recent years. Naturalistic observation, where observation is conducted in the field as opposed to a clinical setting, is considered the gold standard in behavioral data collection. A wide variety of medical and educational professionals rely on video-based behavioral observation systems for their clinical and research work. Most existing systems are difficult to use and do not support any protection in privacy as required by laws. For researchers, VIBE provides an effective, privacy-compliant behavior coding system that can translate to better research outcomes. For therapists, VIBE enhances treatment outcomes by providing an easy-to-use platform to transform therapy videos into training materials for parents to increase their knowledge and involvement. The VIBE system surpasses other state-of-the-arts systems in the following aspects. First, VIBE offers a unique social context preserving privacy sanitization technology that delivers a sweet spot between utility and privacy protection for different users of the system. Second, VIBE includes an innovative algorithmic framework in fusing and calibrating RFID, multiple color-depth cameras and microphones to solve key challenges in voice segmentation, subject identification, behavior tracking and visualization. Third, a number of research prototypes on privacy protected video surveillance systems have appeared in recent years. All of them focus on specific areas of obfuscating and protecting sensitive visual data. VIBE goes far beyond these systems in providing a highly efficient and secure cloud-based privacy information sharing system. Fourth, VIBE supports a wide array of automatic event filters that help customers to identify not only the behavioral episodes, but also the antecedent and the consequence of the behaviors enabling more efficient behavior coding and automatic video summarization. Supported by this award, we have completed our initial customer discovery cycle and have learned the basic principles of running a lean startup. The market opportunity and competitive advantages of the VIBE system have been explored. The attached images showed the distribution of the 100 potential customers who we have interviewed and the business canvas that we have developed. We found that our target customers are researchers and therapists specialized in this area. Many researchers have difficulties in collecting behavior data from schools and community due to stringent privacy rules. Also, they lack an efficient tool to sieve through large video collections. Different from other behavior-coding systems on the market, VIBE offers a novel set of privacy protection tools with full compliance to privacy laws. VIBE also includes event filtering to ease researchers in identifying behaviors of interest. For therapists, they desire better treatment outcomes so as to reduce the number of denials from insurance. As children with developmental disorders require intensive intervention, the best approach to improve treatment outcomes is to provide better training for parents. VIBE offers novel visualization and summarization schemes to preserve important affect cues and social interaction, thereby enabling automatic summarization of a long therapy video into short video clips for parent training. This award has provided financial support for doctoral and post-doctoral students in developing VIBE and resulted in a submission of NSF SBIR phase I proposal.
