Digital Health ' Developments And Challenges

• Published date: 18 January 2021
• Subject Matter: Food, Drugs, Healthcare, Life Sciences, Privacy, Technology, Data Protection, Biotechnology & Nanotechnology, New Technology
• Law Firm: Yigal Arnon & Co
• Author: Ms Yoheved Novogroder-Shoshan, Netanella Treistman and Tamar Tavory

This article explores uses of telemedicine services and AI applications in the health sector, and addresses challenges that are endemic in these technologies.

MORNING, SEPTEMBER 2030

You wake up, it's a regular day. "Mom, I'm not feeling well" yawns the toddler. Hand on forehead. Hmmm... Maybe he's sick. You log into your local telehealth provider, which recommends performing a home swab test. If you don't do it, the provider will automatically inform the toddler's nursery school and he won't be allowed in. Swab the child's throat and process as instructed - the data is on its way to the lab. An alert from the telehealth portal - an artificial intelligence software has analyzed the data and determined that the child is well. Perfect! Off to preschool. In the afternoon the child's temperature rises. The physician's office, automatically alerted by the toddler's wearable bracelet, has already sent an alert that a prescription is on the way. The child's medical insurance has also been notified automatically and sends you a notification that the doctor's prescription will be 85% covered.

While this morning of the future differs significantly from the typical morning in 2020, artificial intelligence (AI), telemedicine and wearables are already integral parts of today's health landscape. By one expert estimate, within ten years the vast majority of surgical procedures will be performed by robots using artificial intelligence and the majority of hospitalizations will be replaced by monitoring and treatment via telemedicine. This article explores uses of telemedicine services, AI applications and wearables in the health sector, and addresses challenges that are endemic in these technologies.

ARTIFICIAL INTELLIGENCE

Artificial intelligence means the performance by a computerized system of tasks that would be considered intelligent were they to be performed by humans.

AI has been implemented in healthcare in patient diagnostic and screening applications, medical data analyses, medical treatment recommendations, pharmaceutical development, and in robots, used to perform surgical procedures. For example, Microbiologists at Beth Israel Deaconess Medical Center demonstrated that AIenhanced microscopes can be used to identify bacteria quickly and accurately. John Hopkins Hospital partnered with GE Healthcare Partners to implement AI-based systems and predictive analyses in order to better manage patient safety, experience, volume, and movement.

AI-enabled wearables and connectivity devices are bringing diagnostic tools to patients, empowering them to manage their health condition AI fertility software and trackers promote women's health, and AI software and bots are used to treat mental health conditions - a field where there is often a scarcity of treatment options

However, AI's special characteristics raise novel challenges. AI can in certain instances lead to output that is at least partially tainted by prejudice, racism, or sexism. AI's algorithms are trained on certain data sets. Where the data set does not represent the totality of the patient population but only a subset thereof, the AI may make false recommendations or predictions when treating a patient from unrepresented group. According to a study published in Nature, widespread racism was revealed in decision making software used by US hospitals.¹ Therefore, careful attention must be paid to the data set input before adopting recommendations based on artificial intelligence, to ensure that discrimination or faulty presumptions do not lead to poor decisions.

In cases where AI serves as a physician decision support tool, questions are raised as to the delineations of the physicians' responsibility and how much weight will be accorded the software's "judgement," at the expense of the physician's own judgement. This is a challenge that will need to be overcome as the field develops.

Special attention is required in a "black-box" situation, where AI tools are employed but one cannot explain the causality between the input and the output - i.e., we are unable to explain how certain input (medical data on a disease for example) leads to a specific output (such as a recommended treatment). Physicians rely on their experience and on medical literature to explain the linkage between the disease and potential treatments. AI may generate insights and achieve better results than any...