The rise of automation and AI has raised fears about job loss, but a team led by Caltech medical engineers suggests that smart tools can help train and assist workers rather than replace them. The team has developed a low-cost device that trains and assists people with minimal laboratory experience to perform complex laboratory procedures. The invention could help address a chronic shortage of trained personnel that currently limits the reach of healthcare in both developing and developed countries.
As a first demonstration of this kind of smart tool, the team developed a new device that monitors a user's progress during sample-pooling, a laboratory process in which biological material from multiple people is combined into a single sample prior to testing. Sample-pooling is a powerful strategy in diagnostic testing because it can cut down on time, cost, and testing supplies. The Caltech team designed the device to guide sample-pooling specifically for nucleic acid amplification tests (NAATs). NAATs are a class of highly sensitive diagnostic tests that are used to detect the genetic material (DNA and RNA) of a wide range of infectious disease-causing agents, including respiratory viruses, the microorganisms that cause sexually transmitted infections, and blood-borne illnesses such as hepatitis C.
The work was led by graduate student Minkyo Lee in the lab of Rustem F. Ismagilov , the Ethel Wilson Bowles Professor of Chemistry and Chemical Engineering, a Merkin Institute Professor, and director of the Jacobs Institute for Molecular Engineering for Medicine at Caltech. Lee, Ismagilov, and their colleagues describe the device in a paper that appears in the journal PNAS Nexus.
"The high cost of NAATs is a critical barrier to healthcare access in global health. Sample-pooling can help, but it requires skills that are often scarce in resource-limited settings," Lee says. "By training users to pool samples reliably, our device addresses two barriers at once: cost and workforce capacity."
Although sample-pooling can reduce per-sample costs and increase testing throughput, the technique is inherently error prone, as workers must transfer clinical specimens correctly, avoid cross-contamination, and maintain accurate documentation. As a result, pooling is often limited to large testing laboratories with access to robotic liquid-handling instruments, such as blood and plasma donor-screening programs that perform infectious disease testing on a large scale.
The Caltech team's new device aims to make pooling more practical in small- to medium-scale health settings, such as mobile or decentralized testing sites that might not have access to trained personnel or expensive automated liquid handlers. The device offers language-agnostic step-by-step instructions and provides clear, real-time feedback. For example, the device alerts the user with sounds and flashing lights if they miss a step in the workflow. The technology helps users fix correctable mistakes and guides them to terminate the process when mistakes are unfixable.
The entire apparatus was specifically designed to use low-cost, off-the-shelf electronic components, 3D-printed modules, and open-source systems.
The authors tested the device with 48 participants, 37 of whom had little or no prior laboratory experience. Compared with paper instructions, the device helped users pool mock clinical samples with higher accuracy, reduce uncorrected handling errors, and acquire volume-transfer skills. Ultimately, when participants used the device they produced more high-quality pools than when they used the paper instructions.
The scientists also validated device performance using archived clinical stool specimens from children in Bangladesh diagnosed with soil-transmitted helminths (STHs), roundworms that cause a group of neglected tropical diseases that currently affect an estimated 1.5 billion people worldwide. STH surveillance exemplifies the challenges the device seeks to address: Massive amounts of STH testing are needed, yet the healthcare training to perform such tests is scarce in the regions that need it most. Device-assisted sample-pooling showed 100 percent agreement with the current leading technique, quantitative PCR (polymerase chain reaction) testing of individual samples for the DNA of an STH-causing parasite.
According to the authors, smart tools such as their training device could logistically and financially enable pooled NAATs for large-scale surveillance programs, including programs that use testing to estimate when the prevalence of diseases is low and can inform decisions about when it is safe to stop administering preventative drugs to the broader population for those diseases.
"AI and robotics are changing how biomedical work is done, but replacing manual workflows is not the only model for innovation," Lee adds. "Technology can also be designed to work with people-helping them acquire the skills needed to perform reliable laboratory workflows in settings where automated systems are not feasible. We hope this work will stimulate development of additional innovative train-and-assist devices that strengthen the workforce, support public health priorities, and expand health care access."
The title of the paper is "A train-and-assist device that upskills novices to strengthen the workforce and expand diagnostic access." Additional Caltech authors of the paper are Xinyue (Penny) Pei, who was a research lab technician in the Ismagilov lab; Si Hyung Jin, senior postdoctoral scholar; Natasha Shelby, scientific research manager; Rani Gera, postdoctoral scholar; Alexander Viloria Winnett (PhD '25), who contributed as a graduate student in the lab, and Colin F. Camerer , the Robert Kirby Professor of Behavioral Economics and the T&C Chen Center for Social and Decision Neuroscience Leadership Chair and director. Mahbubur Rahman of the International Centre for Diarrhoeal Disease Research (icddr,b) in Bangladesh; Nils Pilotte of Quinnipiac University; and Steven A. Williams of Smith College and the University of Massachusetts, Amherst, are also authors of the paper. The work was funded by Caltech's Merkin Institute for Translational Research and the Jacobs Institute for Molecular Engineering for Medicine .
