Written by Anushka Angle

Credit: CIO

Artificial intelligence – AI – has recently been implemented within society, promising to make the day-to-day tasks in a human’s lives much more efficient and accurate. With its algorithm-based programming and mathematical logic, the objectivity of AI might not necessarily be accustomed to the nuanced disparities that exist in our society, as it does not account for empathetic features in its programming. An AI algorithm is developed and trained with data that may embody existing social, racial, and gender biases which could exemplify such prejudice, especially in the medical field where social awareness and empathy are arguably equally as important to quality, effective care as the medical science itself.

Medical Racism Cases

A major issue regarding AI is that mainly due to its novelty, there is insufficient regulation to account for possible problems that may arise. A large number of clinics, along with the general public, are unaware of the aspects of AI, which contributes to a lack of transparency that exacerbates racism in healthcare. In 2019, a bombshell study concluded that a clinical-based algorithm was utilized in many hospitals that decided the prioritization of which patients needed care. The study showed that Black patients were assigned the same level of risk as White patients even though they were deemed sicker than White patients. The authors concluded that this racial bias ultimately reduced the number of Black patients who identified as needing extra care by more than 50%. Since the algorithm used health costs as a substitute for health needs, less money was spent on Black patients who have the same level of need but are deemed as “healthier” than equally sick White patients. If this racial bias was mitigated, the percentage of Black patients receiving extra help would increase from 17.7% to 46.5%. Although this algorithm exposed the existing wealth and income disparities in healthcare and how different races are affected, the incident provokes thought to how AI algorithms are further abetting such medical discrimination.

Another case of AI-worsened bias was exhibited in an algorithm placed in a healthcare facility in Arkansas which determined the number of hours of care the residents of Arkansas would obtain every week if they were to be a patient. There were extreme cuts made to at-home care which unfortunately led to the hospitalization of some patients. Specifically, the new system that was utilized – InteRAI – has a much more objective assessment as compared to the prior human-based system which mainly had occurrences of favoritism and random decisions. However, the bias shown in the online program led the Arkansas Department of Human Services to cite a lawsuit in the state court about the unfairness of how the system determined the medical needs of people with particular disabilities. Arkansas is merely an example of how many states or even other countries are having the same issue where people who really need medical services are not able to easily attain them due to some unfair assessment.

Possibilities for Regulation

The need for quality and equitable healthcare has been a prevalent matter throughout the course of many years to the point where healthcare has become a civil rights issue. The pandemic of COVID-19 has exposed further how the existing socioeconomic inequities and disparities require empathy, which is difficult to be output by an algorithm. In the aftermath of the pandemic, our healthcare system needs to be prioritized. The benefits of AI promise to remove any existing bias from institutions whilst also improving outcomes and diagnoses, but ironically AI has ended up accommodating the existing bias. In order to propose ideas for regulation policy changes among stakeholders, state and federal regulators along with various medical, public health, and clinical groups should work together to address these gaps. A new ACLU paper suggested that demographic information should be required and big corporations such as FDA, HHS, and FTC should establish the best assessment of any differences, such as any racial or ethnic subgroups, in the clearance or approval process.

The Promise of AI

Despite surrounding uncertainty and lack of regulation at the moment, AI has extreme potential to revolutionize the medical field. With its error-free, high-speed algorithms, AI could diagnose conditions more efficiently – crucial to timely life-death cases. The algorithms could also aid in end-to-end drug discovery and development and operate computer systems more effectively than humans. In the form of wearable technology such as limb prosthetics or even devices such as FitBits and smartwatches, AI could allow one to assess their own health and improve self-care. These assessments then mitigate excess workload from healthcare professionals and prevent unnecessary hospital visits. Along with the achieved accuracies said to even exceed human experts, there is fear that it is only a matter of time before certain jobs within the medical field are replaced by AI.

Even though AI has the possibility of depleting or reducing the value of certain medical professions, it “cannot take away the empathy and sympathetic interpersonal care that is provided by a human”. As mentioned before, AI depends on objective aspects in its algorithm and does not take into account the nuances and disparities among people in society, thus exacerbating certain biases. Since interpersonal empathetic communication is equally as important to creating quality, effective care as the medical aspect of healthcare itself, there is no doubt that AI cannot possibly replace many of the professions in the healthcare field.

With the emergence of AI, there is no doubt that it could change the landscape of medicine from a science and efficiency perspective. Instead of treating AI as a possibility for negative change or a factor of opposition, we should treat AI as a tool to augment and improve human capabilities. The benefits of AI, in this way, can be extrapolated to their fullest potential. With any system introduced in society, regulations must be placed to account for any potential harmful consequences. Thus, once regulations have been implemented for AI, the system can then mitigate existing biases instead of promoting them, truly revolutionizing medicine and the world.

Edited by Amna Hassan

---