Houston—A University of Texas System audit criticizing the administration of a $62 million partnership between MD Anderson Cancer Center and IBM Watson does not necessarily mean an end to the effort to treat cancer using artificial intelligence tools at the medical institution.
That’s according to the researcher who led the project at MD Anderson, as well as other UT officials reached by Xconomy.
Karen Adler, a UT System spokeswoman, said UT has an RFP (request for proposal) out for partners that could aid MD Anderson with using the “potential of cognitive computing” for cancer treatment. She added that previous partners would not be excluded, which leaves the possibility open for further IBM involvement.
An IBM spokeswoman wasn’t available for comment.
Houston-based MD Anderson’s Oncology Expert Advisor (OEA) partnership with IBM Watson began in 2012 and was originally aimed at leveraging Watson’s computing power to research the best treatment options, such as clinical trials, for leukemia patients, for example.
But an audit that was conducted last fall and made public January 31 concluded that MD Anderson mishandled aspects of procurement and contract management. These “were not handled in accordance with established policies and procedures,” the report states. “In addition, some actions were taken that are not specific violations of policy or standard procedures but could be perceived as inappropriate or lacking in transparency.”
The UT audit states that IBM pulled its support for the project in September and that the OEA should not be used in clinical treatment of patients.
“If the issue being raised is that OEA is a failure, I disagree,” Lynda Chin, the scientist and researcher who ran the project at MD Anderson, said in an interview. “It is a success technically. It can deliver what we designed it to do. If the question is whether OEA should be an area of investment by the world’s number one cancer center, I’d say, yes. This represents the future of medicine.”
The audit specifically singles out Chin, who is currently the UT System’s vice chancellor for health transformation and chief innovation officer for health affairs, faulting her for signing contracts when she did not have authority to do so, failing to comply with IT policies, and paying invoices even when contracted services were not provided. (She left MD Anderson in April 2015.)
Chin said the change order she signed was “genuinely a mistake,” and one that was quickly noticed and rectified. But she rejected the assertion that the OEA program should be governed as an IT project.
“In my mind from the very beginning of conception and throughout the project, this is an innovation research project,” she said. “It’s an experiment.”
Furthermore, if those were the concerns, IT leaders should have raised their objections, she said. “The CIO, deputy CIO, the CTO, all were involved in our weekly meetings,” she said. “We needed their support and they were engaged.”
She said the finding that invoices were paid even without deliverables being met is “inaccurate.”
“The deliverables were met; the deliverables were reviewed and the invoices paid,” Chin said. “They weren’t exactly identical to how they were written in the contract 18 months prior. That’s simply the nature of innovation projects.”
In his response to the audit, UT Chancellor William McRaven expressed that he understood the view that projects like OEA could be considered research and development work, though he also wrote that in the future, MD Anderson “must improve its record systems to support payments to contractors on projects that evolve over time.”
Chin says she does not know the next steps for the OEA project, as she has not been involved with it since her departure in 2015.
In a message to MD Anderson employees this week, CEO Ron DePinho expressed support for what he called “pioneering accomplishments” in the OEA project. (Chin is married to DePinho.)
“The project was placed on hold to allow us to develop a more mature commercialization plan for a cognitive system to aid other cancer care teams in applying advanced knowledge for patients’ benefit,” he wrote. “We believe cognitive computing has the potential for tremendous applications in cancer.”
