Google Contact Lens

~ the last words ~

On 16 January 2014 Google X put up a post about a contact lens that would measure the glucose in your tears. Brian Otis and Babak Parviz wrote it, both of them out of the electrical engineering faculty at the University of Washington, and the device they described was very small.

A wireless chip and a miniaturized glucose sensor sat between two layers of soft lens material. A pinhole let tear fluid seep through to the sensor. The whole thing was powered and read over an antenna thinner than a human hair, and the prototype produced a reading once a second.

The sensors sat outside the pupil and the iris, kept clear of the parts of the eye that are easy to damage. There were plans for tiny LEDs in the lens to warn the wearer when their glucose crossed a threshold, which is the part everybody quoted, because it meant a diabetic could be told something by their own eye instead of by a finger prick and a strip.

Novartis bought in fast. On 15 July 2014 its Alcon eye-care unit licensed the technology and took on developing it into something a patient could be given. Google's life sciences arm became Verily in December 2015 and the program went with it. My own working days go into sensors bolted to small flying machines, which is why this failure reads as familiar rather than exotic. Underneath a project like this there is always a quiet assumption that one measurable quantity will stand in for the different one you actually care about. The electronics were the impressive half here and they were never where the risk sat.

That assumption is what failed. On 16 November 2018 Otis, by then Verily's chief technical officer, posted that the clinical work had shown "insufficient consistency in our measurements of the correlation between tear glucose and blood glucose concentrations to support the requirements of a medical device." He gave two reasons: other biomolecules in tears interfered with the reading, and the tear film would not hold still enough to be measured reliably.

~ the record ~

~ a doubt that was published first ~

The correlation problem was already in the literature in 2014. Studies had found tear glucose tracking blood glucose only weakly, nowhere near the accuracy a glucose meter has to hit, and outside researchers said so publicly while the press cycle was still running. Four years and a Novartis licensing deal went past between that doubt and the answer.

Verily and Alcon kept working together on the other lenses, the accommodating one for presbyopia and the intraocular one for cataract surgery. Those had nothing to do with blood chemistry, which is why they survived the same announcement that killed the glucose version.

~ what shipped ~

Nothing. Nobody ever wore one of these outside a study, and no model number for it exists anywhere in the world. What four years produced was a licensing agreement and one unusually honest blog post about why a measurement does not work.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »