Enhanced Communication for ALS Patients; IBM Watson for Oncology Repurposed

Artificial intelligence (AI) is a ubiquitous technology and field of study that is taking industries by storm. In healthcare, AI is applied to aid physicians in diagnosing and treating patients. As a biomedical engineering student, I sought out to research and develop AI applications in healthcare. In my technical project, I worked with three classmates to build a blink-detecting camera that enables amyotrophic lateral sclerosis (ALS) patients to communicate with their caregivers. In my STS project, I examined IBM Watson for Oncology, a notable AI cancer diagnosis tool’s contingent use and acceptance by patients, clinicians, and hospitals, as a complementary aid of, rather than a crutch for, physician expertise, diverging from IBM’s original vision. Through these two projects, I expanded not only my technical understanding of how AI can be harnessed to improve patient care, but also my social understanding of the reality that a technology’s success and ultimate purpose will be determined by its user. ALS is a neurodegenerative disorder, causing progressive paralysis and loss of motor function. Late-stage ALS patients have impaired breathing and speaking function. The loss of communication severely debilitates a person’s self-esteem and quality of life and increases caregiver burden due to 24/7 supervision. In my Technical Report, I detail my capstone project aimed at developing a novel Augmentative/Alternative Communication device for ALS patients that harnesses patient blinks for communication. We designed and built a camera attached to ALS patient bilevel airway pressure masks integrated with a machine learning-powered algorithm to detect blinks. In the future, the camera device will be iterated to convert blink combinations to communication signals for the caregiver and will be tested with non-ALS and ALS subjects (ongoing IRB application). In my STS Research Paper, I examine the sociotechnical dimensions of IBM Watson for Oncology’s failed integration into healthcare. Critics argue that Watson failed as an automated diagnostics tool for physicians as it often provided inaccurate treatment recommendations, attributing limited training data and patient complexity as root causes. However, I argue that although IBM failed in its original intention to provide accurate diagnosis and treatment recommendations—almost replacing the physician—the initiative succeeded in ways that patients, physicians, and hospitals would accept such a device: a complementary resource for evidence and data, rather than a replacement of physician expertise. Using the sociotechnical frameworks Social Construction of Technology and User Agency, I examine how Watson’s users are the ultimate metric of its success. My sociotechnical analysis of IBM Watson has broadened the approach I take to the development of my ALS communication device. As my group finalizes its prototype for human testing, I am aware that the ultimate test of success and source of future direction will be the ALS patients themselves. Like Watson, there are innumerable insights from the target users of our device that we would not guess ourselves. I now value human subject testing and user feedback even more and have learned to seek it much earlier in future design projects.