As hospitals adapt to various viral challenges, a new kind of viral phenomenon is gaining attention: generative artificial intelligence (AI) in the workplace. Prominent healthcare institutions, like Boston Children’s Hospital, are at the forefront of embracing technological advancements to revolutionize productivity in the healthcare industry.

With healthcare accounting for approximately 18 percent of the U.S. GDP, it is natural for organizations to seek the latest technologies to enhance their operations. Boston Children’s Hospital, known for its consistent ranking among the top children’s hospitals in the United States, has a dedicated “Chief Innovation Officer” named John Brownstein. Brownstein, an epidemiologist, oversees the Innovation & Digital Health Accelerator division, which combines technology and health to drive innovation.

Brownstein has previously developed platforms such as “Flu Near You,” repurposed as “Covid Near You” during the early days of the pandemic, and now exists in a more general form as “Outbreaks Near Me.” These platforms effectively track pathogens. Currently, Brownstein is directing his focus towards AI.

According to Brownstein, AI should be seen as an augmentation rather than a replacement for human involvement in healthcare. He emphasizes the importance of human involvement throughout the AI-enabled processes. For example, envisioning a machine kiosk in hospitals where patients can receive answers to various questions, Brownstein proposes using AI to enhance patient experiences, even addressing inquiries like “Where can I pray?” This approach aims to provide efficient and personalized healthcare services while recognizing that not everyone is inclined to engage with others.

However, the involvement of AI in processing actual patient data raises concerns. Mildred Cho, a professor of pediatrics at Stanford’s Center for Biomedical Ethics, points out that the qualifications of prompt engineers primarily focus on computer science and coding expertise, rather than a nuanced understanding of medicine and healthcare systems. Cho highlights the potential risks of flawed assumptions and the inadvertent incorporation of biases or errors into AI systems built on imperfect data.

Acknowledging these concerns, Brownstein assures that the prompt engineers working with his team are not isolated but actively consider the implications of imperfect data. While large language models often rely on “putting in a bunch of data and hoping for the best,” Brownstein’s team devotes time to address data imperfections and ensures a thoughtful approach to AI implementation in healthcare.

One of Brownstein’s intriguing visions for the future involves AI-driven customization of discharge instructions. Instead of generic instructions, AI algorithms trained on individual patient information can provide tailored recommendations. For instance, the system can suggest where to buy specific medication based on the patient’s location or modify instructions based on the patient’s age and other attributes, using the most compelling voice to enhance adherence to the instructions.

While some experts, like David Himmelstein, a professor at the CUNY School of Public Health, view AI-powered discharge instructions as a positive development, concerns about privacy arise. Himmelstein questions who will have access to patient information and raises the issue of data ownership if companies like Microsoft or Google are involved. However, Boston Children’s Hospital aims to develop internal AI systems to keep patient data within the hospital’s walls, mitigating privacy concerns.

Himmelstein also highlights that automation in hospitals is not a new concept and has not necessarily led to bureaucracy-free environments. He refers to a historical IBM video from 1961 that promised electronic systems to eliminate errors and reduce bureaucracy, reminding us that achieving seamless efficiency in healthcare has been an ongoing endeavor.

As Boston Children’s Hospital explores the potential of generative AI in healthcare, considerations around privacy, data accuracy, and human involvement remain vital. The hospital’s commitment to developing internal AI capabilities signifies a cautious and patient-centered approach, ensuring that the benefits of AI are harnessed while maintaining ethical and privacy standards in the healthcare sector.

By Impact Lab