TL;DR:
- AI tools in NYC hospitals swiftly assess patient conditions, such as malnutrition, delirium, and ICU admission, before doctors enter the room.
- Administrative tasks like appointment booking and prescription refills are streamlined through AI systems.
- AI programs provide tailored pregnancy advice and can identify signs of diseases like breast cancer.
- AI prioritizes specialists’ time, guides medical decision-making, and detects early signs of critical conditions.
- Speed and efficiency are the key advantages of AI tools, freeing healthcare professionals to focus on patient care.
- AI assists in targeting at-risk patients, leading to faster treatments and recoveries.
- AI algorithms identify patients in need of time-sensitive procedures, such as clot removal for stroke patients.
- AI-enhanced chatbots facilitate patient interactions and provide medical guidance.
- Ethical considerations and rigorous testing are paramount in the deployment of AI tools.
Main AI News:
The transformative potential of artificial intelligence (AI) has extended beyond the realms of learning and work, making significant strides in the medical field. However, many New Yorkers remain unaware that these innovative algorithms are already revolutionizing life-saving decisions in hospitals on a daily basis.
Within seconds, AI tools at three prominent New York City hospital systems can accurately assess a patient’s probability of malnutrition, delirium, intensive care unit (ICU) admission, and even mortality, long before a physician steps foot into the hospital room. While physical examinations are not within AI’s current capabilities, these systems assist in streamlining medical decision-making, prioritizing specialists’ time, and detecting early signs of critical conditions like breast cancer. The invaluable role of AI in healthcare was the focal point of a recent summit organized by the New York Academy of Sciences, where clinicians expounded upon its applications.
Sara Wilson, Senior Director of Clinical Nutrition Services at Mount Sinai Health System, attests to AI’s ability to identify patients who may have otherwise gone unnoticed. With its unparalleled speed and comprehensive analysis of various metrics, AI surpasses human capabilities, allowing medical professionals to focus on direct patient care. Wilson specifically highlighted an AI tool that identifies individuals at risk of malnutrition. This predictive model analyzes patient charts, emphasizing factors that may predispose them to undernourishment, such as weight and bloodwork. If a patient’s risk factors surpass a certain threshold, the system promptly alerts them for a follow-up visit with a registered dietitian.
The remarkable advantage of AI tools, as explained by hospital staff, lies in their rapid data processing abilities. While it would take a human considerable time to sift through medical records, AI algorithms accomplish this task in a fraction of the time. Consequently, doctors and nurses can allocate more time to patient interactions, which enhances the quality of care provided.
Additionally, the expedited identification and treatment of high-risk patients are critical in conditions like delirium, particularly for older individuals. Dr. Joseph Friedman from Mount Sinai Health System elucidated that before employing their AI model, the hospital faced challenges in promptly engaging in aggressive and decisive treatment due to the time-consuming nature of assessments. However, with AI support, healthcare professionals can target at-risk patients more efficiently, leading to faster recoveries. Previously, clinicians would need to screen up to 100 individuals daily to identify only two delirious patients. With the AI program in place, as many as ten patients can now be identified for every 25 screened each day.
Moreover, AI tools have proven instrumental in identifying patients requiring time-sensitive, life-saving procedures. New York City Health + Hospitals utilizes an algorithm to examine brain scans of stroke patients, swiftly flagging suitable candidates for clot removal surgery. NYU Langone employs a care planning tool that estimates a patient’s likelihood of dying within the next two months, enabling medical professionals to provide tailored care accordingly.
In addition to medical care, some hospitals are exploring the utilization of chatbots to engage with patients. These chatbots serve a range of purposes, from assisting with everyday tasks like scheduling appointments to delivering medical advice. For instance, Mount Sinai employs a chatbot that guides anxious patients in deciding whether to schedule a regular doctor’s appointment, visit an urgent care facility, or proceed to an emergency room. Similarly, Northwell Health introduced an AI-enhanced pregnancy chat app, providing personalized advice while screening for common symptoms. The app also facilitates connections to healthcare workers or prompts users to call 911 if their responses indicate emergencies, such as preeclampsia or self-harm ideation.
Enthusiasm resonates among hospital staff interviewed by Gothamist, who recognize the potential of advanced language models like ChatGPT to generate text in response to customized prompts. These systems could assist healthcare providers in writing tasks, including patient communication.
Nevertheless, it is crucial to acknowledge that AI is not infallible. A tool developed to predict the deadly infection response called sepsis required extensive redesigning following the identification of flaws by researchers and journalists. Furthermore, AI can inadvertently reproduce biases present in its human creators, as exemplified by NYC Health + Hospitals’ decision to forgo a tool predicting the likelihood of patients missing scheduled appointments due to concerns about bias.
Healthcare institutions, including New York City Health + Hospitals, Mount Sinai Health System, and NYU Langone Health, emphasize the importance of ethics committees and rigorous testing to ensure the reliability and accuracy of their AI tools. This is particularly significant for tools that predict patient decline or mortality. A care planning tool estimating a patient’s odds of dying within the next two months necessitates extensive discussion and encourages doctors to engage in conversations with high-risk patients regarding end-of-life preferences, which can then be documented in living wills or other instructions.
Representatives from these institutions assert that AI tools only inform medical professionals’ decision-making processes and do not replace them. Human involvement remains paramount, as the models serve as one aspect among numerous data points considered by healthcare providers.
Despite the pervasive hype and predictions regarding AI’s encroachment on medical careers, doctors and clinicians remain unconcerned about AI superseding their roles, especially when physical interaction is imperative. Dr. Friedman aptly captures this sentiment, stating, “The AI serves us. We don’t serve it.” Comparing it to a Star Trek holographic doctor making decisions, he reassures that such a scenario is yet to materialize.
Conclusion:
The utilization of artificial intelligence in New York City hospitals is transforming patient care. The integration of AI tools has improved the speed and accuracy of diagnosing conditions, prioritizing treatment for at-risk patients, and assisting in medical decision-making. By automating administrative tasks and providing personalized guidance, AI enhances overall healthcare efficiency. However, ensuring ethical implementation and rigorous testing remains crucial to address potential biases and maintain patient trust. The market for AI in healthcare is poised for growth as hospitals recognize its immense value in augmenting healthcare professionals and improving patient outcomes.
