Material Gains
Applying Principles of Industrial Automation in Healthcare Can Deliver Better Services for All
How AI and lab-on-a-chip pave the way for economical solutions.
We could be moving into the end game with the Covid-19 pandemic, at least as far as the severest effects are concerned. Clearly, the virus and its mutations are here to stay, and the future will be about protecting us through immunization and developing better treatments. The fact that effective vaccinations have become available only a year after the pandemic was recognized is remarkable. It’s partly due to the speed with which researchers have been able to do the data crunching needed to model and understand how best to attack the virus.

In the past, the computations involved in sequencing the virus DNA would have taken vast quantities of computer time and prolonged development of the vaccine. Cloud computing using AI accelerators has dramatically shortened the time to complete the technical work involved in creating the vaccines now being rolled out.

It would be great if we could harness our technologies to create an early warning system when clusters of unusual diseases or events occur anywhere in the world. That’s exactly what organizations like BlueDot are doing right now. Indeed, BlueDot says it spotted the cluster of unusual pneumonia cases in Wuhan in December 2019 that we now know was the coronavirus. To monitor the spread of infectious diseases around the world, it analyzes a huge number of variables, not only official public health data but also climate information, international travel patterns, animal and insect population data, and others. This relies on the ability of AI to detect patterns, and exceptions to those patterns, hidden within the enormous body of information. By sifting through the reports and data points collected every few minutes, 24 hours a day, from sources around the world, using techniques like machine learning and natural language processing, BlueDot brings a small number of cases to the attention of experts for further investigation. Only with AI do we have a hope of finding those cases.

sometimes gets the
right answers
for the
wrong reasons
when training, which could
produce machines
that make
incorrect inferences.
While AI empowers us to monitor the effects of variables such as climate and movement of people, and disease carriers such as insects, with greater precision than ever before, capturing raw data that accurately describe the condition of patients calls for large numbers of sensors that are easy to use and inexpensive. The University of Bath in the UK is leading a project to tackle an emerging diabetes epidemic in Turkey, for which it has developed children-friendly patches for painless glucose quantification. The patches are designed to be economical because typical noninvasive sensors that do not require finger pricking are simply too expensive for large-scale studies in developing countries. They contain an array of hydrogel microneedles that painlessly capture subcutaneous interstitial fluid to be tested using a µTAS (Micro Total Analysis System) sensor platform fabricated on a flexible printed circuit, using special substrate materials for the carrier for the lab-on-a-chip components. Optimizing the properties depends not only on the right amount of flexibility but also biocompatibility.

In the industrial world, cost savings are achieved through intensively automating processes to reduce the cost of labor and increase output. Accuracy and precision are also improved. Automation can bring these advantages to healthcare, too, enabling more people to benefit from better services and enjoy better patient outcomes. We have seen how human-guided surgical robots have improved on the fine-motor skills and visual acuity of human surgeons. Researchers have also successfully used augmented reality in surgery to overlay images such as CT scans in the field-of-view that show the locations of items such as bones and blood vessels to help direct procedures such as reconstructive surgery or neurosurgery. In addition, we are seeing precision optical technology from the PCB inspection business now being applied in the medical domain to help increase the precision of brain surgery, permitting smaller incisions that minimize trauma and enable faster recovery times.

Full automation is now a promising next step. I’m involved with a group that fundraises for cataract surgery in India. India has high instances of “avoidable blindness” simply because many people are too poor to afford cataract surgery. The undersupply of qualified surgeons and costs of training are key challenges we face. We are optimistic automated laser surgery can help address these issues by reducing reliance on training and increasing the number of procedures that can be carried out per day.

Historically, laser surgery has been seen as the premium option, offering lower risk and better outcomes for the few. With greater automation, there is hope the benefits could ensure affordability for many and increase the number of operations per year. Likewise, I see full automation as the way forward for many surgical procedures, most likely using single-purpose machines. It could be a few generations before a general-purpose diagnosis and surgical station like the MedPod 720i imagined in the movie Prometheus becomes not only technically feasible but affordable.

There may, of course, be fears for the future. I think both human surgeons and surgical machines will be needed for a long time to come, due to generally high demand. Moreover, human expertise will be essential to direct services and oversee individual cases. As AIX-COVNET, a project developing machine learning to analyze lung x-rays from Covid-19 patients, noted, AIs will sometimes get the right answers for the wrong reasons when training, which could produce machines that make incorrect inferences. We need not only data scientists, but also domain experts, to create the machines that will provide these much needed services.

Alun Morgan smiling
Alun Morgan
is technology ambassador at Ventec International Group (;