IBM Watson Layoffs: Is Healthcare AI Ready for Prime Time?
IBM recently reported a round of layoffs at its Watson Health unit, which uses artificial intelligence (AI) to work in healthcare. The company spokesman, Doug Shelton, told The Herald Sun, that the layoffs “affects a small percentage of our global Watson Health workforce as we move to more technology-intensive offerings, simplified processes and automation to drive speed.”
MedCityNews questions this, pointing out that prior to the Memorial Day weekend, The Register reported the company laid off 50 to 70 percent of Watson staff members in Dallas; Ann Arbor, Michigan; Cleveland; Denver and other locations. At least part of the underlying issues appears to be whether the hype behind AI in healthcare was supported by practical uses. MedCityNews writes, “A 2017 interview with a former IBM employee who worked in the company’s life sciences group showed the hype around AI’s impact on healthcare was also felt internally at the company. He explained that though marketing budgets were large, the talk never materialized into a tangible off-the-shelf product.”
These questions are underlined by a breakup last year between IBM Watson and MD Anderson Cancer Center. The two had worked together since 2013, where MD Anderson stated it “is using the IBM Watson cognitive computing system for its mission to eradicate cancer.” But in late 2016 the project was put on hold and auditors at the University of Texas in a report aid the project cost MD Anderson more than $62 million but didn’t meet its goals.
A STAT report suggests the problems with IBM Watson had less to do with whether it was effective and more about internal rivalries and a lack of organization.
Although IBM Watson was probably the best known in this space, it’s not the only one, of course. Google and Amazon are both heavily involved in AI, healthcare and analytics.
A study was recently published in the Annals of Oncology describing a Google deep learning convolutional neural network (CNN) that was trained to identify skin cancer. Lead author Holger Haenssle told FirstWordPharma, that to train Google’s Inception v4 CNN architecture, “we showed [it] more than 100,000 images of malignant and benign skin cancers and moles and indicated the diagnosis for each image.” He said that “with each training image, the CNN improved its ability to differentiate between benign and malignant lesions.”
The AI’s performance was compared to those of 58 dermatologists. About half had more than five years of experience, 19 had between two and five years, and 29 percent had less than two years of experience. The dermatologists were first shown just the dermoscopic images (level 1) and asked to diagnose either melanoma or benign nevus, as well as indicate how they would manage their diagnosis. Four weeks later, they were given a similar test, but also were provided clinical data about the patient and close-up images of the same cases (level II).
Compared to the AI, at level one, the doctors had a mean sensitivity and specificity for lesion classification of 86.8 percent and 71.3 percent, respectively, but were “significantly outperformed” by the CNN AI. At level II, the dermatologists did a little better, with a mean sensitivity of 88.9 percent and specificity of 75.7 percent. Still, the CNN AI had a specificity of 82.5 percent.
“Our data clearly show that a CNN algorithm may be a suitable tool to aid physicians in melanoma detection irrespective of their individual level of experience and training,” the authors of the study wrote. But they also pointed out that at level I, 13 of the 58 dermatologists “showed a slightly higher diagnostic performance than the CNN.”
On June 11, Saama Technologies, a data analytics company in Campbell, California, announced it had launched a set of AI capabilities designed to help in drug development. The system, DaLIA (Deep Learning Intelligent Assistant), “harnesses Natural Language Processing (NLP) and Natural Language Understanding (NLU) to facilitate an unprecedented conversational experience with clinical trial data that will overcome obstacles historically associated with clinical development, and improve the life sciences industry’s ability to deliver safe and effective therapies.”
And although it may not be completely clear how much AI is going to be used in making healthcare decisions any time soon, it’s fairly clear that biopharma companies are utilizing AI in sorting through the massive reams of data they work with.
On June 12, Juvenescence, based in Douglas, Isle of Man, closed on an oversubscribed Series A financing worth $50 million. The company is focused on developing therapeutics for longevity, and leans heavily on artificial intelligence, specifically machine and deep learning through its part ownership of Insilico Medicine, and its ownership in NetraPharma and Juvenescence AI.
Although it’s probably not wise to completely rule out IBM as a player in the field, others point to Google and other Silicon Valley companies as the likely dominant figures in AI and healthcare. Chamath Palihapitiya, chief executive officer and founder of Social Capital, a Palo Alto, California-based venture capital company, told CNBC’s “Closing Bell” in 2017, “Watson is a joke, just to be completely honest.” He said Google and Amazon are “head and shoulders ahead of every single other company. The companies that are advancing machine learning and AI don’t brand it with some nominally specious name that’s named after a Sherlock Holmes character.”
IBM’s Watson is actually named after Thomas J. Watson, founder of IBM. But Palihapitiya went on to say, “What you do when you innovate in machine learning and artificial intelligence is you spend enormous amounts of time collecting enormous amounts of data. I think what IBM is excellent at is using their sales and marketing infrastructure to convince people who have asymmetrically less knowledge to pay for something.”