Lawsuit Claims the Mayo Clinic's Use of AI Is Butchering Patient Care
Source: Futurism
In a new civil suit, former Mayo Clinic research director and AI compliance lead Traci Tamiko Eto alleges that the globally recognized hospital retaliated against her for blowing the whistle on its disastrous implementations of AI health tools.
According to reporting by Minnesota Public Radio, Eto first joined the Mayo Clinic in 2023. After flagging potential privacy problems linked with the hospitals Mayo Clinic Platform, an AI-integrated data system, Eto says her supervisor ignored her concerns. Instead, she says her boss insisted that fixing the issue would jeopardize the pace on ongoing research projects, which in turn would compromise Mayos competitive advantage.
Eto continued, MPR reports, flagging multiple failures of Mayo Clinic higher-ups to follow federal regulations on review processes of new tech, like MAYA, the clinics AI-integrated digital assistant. According to the lawsuit, the team working on MAYA deleted unflattering test results, mischaracterized the tools abilities, and made decisions that put data security in jeopardy.
Tucked inside her ten separate whistleblower complaints, MPR notes, is the allegation that the team working on MAYA knew the tool had an error rate as high as 67 percent. Eto alleges that instead of flagging that horrifying number, MAYA staffers tried to hide it.
-snip-
Read more: https://futurism.com/health-medicine/lawsuit-mayo-clinic-ai-tools-hospital-maya
From a separate story by a Star Tribune writer:
The lawsuit, filed Monday in U.S. District Court in Minnesota, accuses Eto's superiors of manipulating the oversight process for medical research, which typically relies on independent institutional review boards (IRBs) to set the conditions by which studies can take place safely and ethically. Having an independent entity make those decisions is supposed to remove bias from research studies, but the lawsuit claims Mayo leaders in some cases pressured IRBs to approve studies, steered research proposals to IRBs that aren't as stringent, or just bypassed IRBs and green-lighted research on their own.
-snip-
The health system has positioned itself as a global leader in the development of AI technologies and tools to guide health care decision-making. A conference last month featured numerous studies by Mayo researchers about whether AI tools could predict cognitive declines in patients, assess the severity of head and neck tumors, or even predict tantrums in children with behavior disorders.
-snip-
Eto suffered depression and needed time off as a result of the work stress, the lawsuit alleges. When she returned to work, she was given weeks to find another role at Mayo or be terminated. The lawsuit also alleges Eto was largely responsible for the creation of one AI tool, but that Mayo de-emphasized her role in a patent application in a way that could limit her earnings from its commercial usage.
Specific allegations in the lawsuit include that Mayo researchers tried to cover up or ignore research results showing that the health system's MAYA digital assistant tool had a 67 percent error rate. In a separate situation, Eto claimed she opposed the use of a high-risk investigational medical device for a cardiac surgery because it hadn't been properly vetted by an IRB; she said her supervisor ignored her objection and approved the use of the AI-enabled device for a surgery in another country.
-snip-
harumph
(3,635 posts)ultralite001
(2,888 posts)for his dementia + other health issues...
This development is 💔💔💔...
No individual's quality of life or health care should be jeopardized by untested, unvetted,
unverified, experimental AI programming...
Keep medicine in the hands of PROFESSIONALS... + Let the PROFESSIONALS
practice medicine... Not insurance companies + / or pharmacy benefit managers...
ultralite001
(2,888 posts)Bayard
(30,908 posts)ultralite001
(2,888 posts)TIA
paleotn
(23,275 posts)Probably recommends ivermectin for damn near everything.
Until they can insert a digital connection into experienced medical professional's ears and download a bazzilion collective years of experience, and more importantly, how to use that experience, it will be nothing but a very expensive "well duh / that's not actually true" tool. It's not just the information that's important. It's how it's processed and applied to a nearly infinite number of circumstances. Humans can do that. Machines not so much. They can help with the easy stuff. There isn't all that much easy stuff.
reACTIONary
(7,484 posts).. . Trained on public data like Facebook and Redit. The lawsuit mentions two systems: Mayo Clinic Platform (MCP) and MAYA Digital Assistant.
MCP is divided into three functions: access to de-identified patient data and medical records for AI training,; evaluating digital health solutions for fairness and accuracy; and a global, decentralized data network linking healthcare organizations.
The complaint regarding MCP was that there were problems with the de-identification of the data. While this is a serious problem it isn't specifically AI related and it does not relate to making specific diagnostic or treatment decisions.
MAYA is an AI-integrated digital assistant developed at Mayo Clinic designed to enhance data retrieval. It assists Mayo Clinic employees in accessing the Electronic Medical Record (EMR) and intranet systems to query clinical and administrative information. It seems to have a summarization capability, and according to the lawsuit had a high error rate that was being covered up.
Erroneous summaries could lead to errors in diagnostic or treatment decisions, so this is a serious concern, but once again, the system does not seem to be making, on its own, specific diagnostic or treatment decisions.
warmfeet
(3,365 posts)use AI.
flashman13
(2,688 posts)GiqueCee
(5,265 posts)... to make medical decisions... well, "Houston, we have a problem."
paleotn
(23,275 posts)Vastly worse.