As the Chief Compliance Officer and Institutional Official of Advarra, the second largest company in the US specializing in institutional review board (IRB) and other regulatory consulting services, part of Michele Russell-Einhorn’s job is to review clinical trials and determine whether they are ethically adequate to be carried out.
Recently, she came across several projects that require participants to pay enrolling fees. “Some plan to ask participants to pay $7000 or so… Another wanted to ask for upwards of $250,000… There were serious concerns about how ethical it was to charge people to participate in the research – and whether it was absolutely necessary,” Russell-Einhorn told STAT.
Most clinical trials are done voluntarily with participants randomly assigned to the experimental group – to be given the intended medication or medical procedure, or the control group – which will receive a placebo. The process tends to be single-blind (i.e., participants unaware of which group they are in) or double-blind (i.e., both researchers and participants do not know which group they are in), to ensure any capture effect is solely a result of the medication or medical procedure.
Usually, the US Food and Drug Administration (FDA) will
Apart from patients’ welfare, the authority is also concerned if these fee-paying trials are enterprises’ disguise to look for profit from desperate patients as it conforms a common misconception that clinical trials equate to a new form of treatment that renders new hope. Besides, accepting only fee-paying participants may generate results that do not represent the majority or a diverse population. Since there is an absence of a control group, it will be unclear if the eventual result comes from the medication or medical procedure or something else.
An important lesson for AI
Unlike clinical trials, there is no lack of funding for artificial intelligence (AI) research. Based on a new report by CB Insights, funding for healthcare-related AI has reached new highs. In the second quarter of 2019, healthcare AI startups had received $864 million in venture capital, more than the $764 million earned around the same period last year.
However, AI also faces a lack of good data. It has been reported some startups are turning to fabricate data to train their algorithms. As competitions between companies get fierce and an AI regulatory framework is generally absent, there are relatively more loopholes that one can tap onto.
AI has to undergo validation tests to demonstrate its safety and credibility, it will not be a surprise if companies reuse some of the training data or employ similar data to prove the algorithm works. The personalized nature of AI could mean patients only those who can afford will have access to the technology, while others may have to stick to traditional modes.
The SACHRP committee is still finalizing its recommendations, it will be at least a year before we get to know if such fee-paying clinical trial would be encouraged. Meanwhile, it is the patients themselves who have to decide if they will like to pay for something novel but unknown.
A science writer with data background and an interest in the current affair, culture,