When Ama­zon de­vel­oped a ma­chine-learn­ing tool to make its re­cruit­ment process more ef­fi­cient — the man-made sys­tem ab­sorbed the gen­der-bias of its hu­man mak­ers, and the project was abort­ed. In the field of bio­phar­ma­ceu­ti­cals, the way re­searchers train their ma­chine learn­ing al­go­rithms can skew the out­come of pre­dic­tions. But be­fore those pre­dic­tions can be made, the en­gine must learn to read to make sense of ex­plo­sive vol­ume of knowl­edge out there.

Yian­nis Ki­a­chopou­los

That is what Lon­don-based Causaly has set out to do, and has raised $4.8 mil­lion as it works on re­fin­ing its tech­nol­o­gy, co-founder Yian­nis Ki­a­chopou­los told End­points News in an in­ter­view.

The key is­sue is find­ing ev­i­dence in the first place be­cause 90% to 95% of search re­sults are noise, he said. “The prob­lem that we’re solv­ing is find­ing the ev­i­dence in the first place. The read­ing part and the judg­ment part is still up to the hu­man.”

Mil­lions of bio­med­ical ar­ti­cles have been pub­lished so far, and thou­sands are added each month. The tiny 11-per­son start­up Causaly, which is work­ing with No­var­tis and a host of oth­er bio­phar­ma­ceu­ti­cal com­pa­nies, has de­vel­oped ar­ti­fi­cial in­tel­li­gence tech­nol­o­gy that process­es lan­guage from this avalanche of pub­lished bio­med­ical da­ta — and is de­signed to ex­tract causal re­la­tion­ships the way hu­mans can — ex­cept faster and more ef­fi­cient­ly.

The tech­nol­o­gy is de­signed to iso­late rel­e­vant da­ta and make it vi­su­al­ly ac­ces­si­ble and that’s what sci­en­tists want, Ki­a­chopou­los said. “Our users are specif­i­cal­ly ask­ing us not to make these judg­ments for them…and that’s how we see our­selves as aug­ment­ing hu­mans and not kind of re­plac­ing judg­ments or any­thing like that, that gets more prob­lem­at­ic.”

Ar­tur Saud­abayev

Users are giv­en the op­tion of us­ing fil­ters, for in­stance, they can iso­late da­ta that em­anates on­ly from ran­dom­ized clin­i­cal tri­als or from jour­nals with a de­fined ‘im­pact fac­tor’ or ar­ti­cles that have been pub­lished dur­ing a spe­cif­ic time pe­ri­od. The plat­form is not judg­ing whether that ev­i­dence is cred­i­ble, he said. “We leave the judg­ment to the hu­man, but we give the hu­man the tools to make the judg­ment in a bet­ter way.”

There is an­oth­er lay­er of bias en­trenched in da­ta — and that is lin­guis­tic bias. Nei­ther Causaly’s tech­nol­o­gy nor hu­mans can ful­ly elim­i­nate that be­cause cul­tur­al con­texts play a key role in the way da­ta is processed and ar­tic­u­lat­ed, he added.

Causaly raised $1 mil­lion in seed fund­ing last year, and this in­jec­tion of se­ries A cap­i­tal was led by Pen­tech and EBRD Ven­ture Cap­i­tal, with par­tic­i­pa­tion from ex­ist­ing in­vestors, in­clud­ing Marathon Ven­ture Cap­i­tal.

It is hard­ly the first AI com­pa­ny at­tempt­ing to as­sist de­ci­sion-mak­ers in the fields of phar­ma­ceu­ti­cals and health care to ag­gre­gate and syn­the­size in­for­ma­tion. A grow­ing list of star­tups in­clud­ing Am­plion, Biore­late, Da­ta4Cure, Ev­id Sci­ence, Inno­plexus, In­ve­ni­AI, Lin­gua­mat­ics, Meta, Plex Re­search, Quer­tle, Re­search­ably and nfer­ence are all work­ing on ways to help sci­en­tists and re­searchers di­gest and make in­fer­ences from the del­uge of bi­o­log­i­cal da­ta on of­fer to­day.

Mean­while the tra­di­tion­al, and large­ly free, search tools haven’t ex­act­ly be­come ob­so­lete. “When I talk to our cus­tomers, our biggest com­peti­tor is the sta­tus quo,” not­ed Ki­a­chopou­los. “And that is Google Schol­ar, PubMed.”

In a bewildering twist to the long-suffering market for antibiotics — there has actually been a bidding war for an antibiotic company: Tetraphase.

It all started back in March, when the maker of Xerava (an FDA approved therapy for complicated intra-abdominal infections) said it had received an offer from AcelRx for an all-stock deal valued at $14.4 million.

The offer was well-timed. Xerava was approved in 2018, four years after Tetraphase posted its first batch of pivotal trial data, and sales were nowhere near where they needed to be in order for the company to keep its head above water.

Lonza chairman Albert Baehny took his time headhunting a new CEO for the company, making it absolutely clear he wanted a Big Pharma or biotech CEO with a good long track record in the business for the top spot. In the end, he went with the gold standard, turning to Roche’s ranks to recruit Pierre-Alain Ruffieux for the job.

Ruffieux, a member of the pharma leadership team at Roche, spent close to 5 years at the company. But like a small army of manufacturing execs, he gained much of his experience at the other Big Pharma in Basel, remaining at Novartis for 12 years before expanding his horizons.

Another Big Pharma is entering the Covid-19 antibody hunt.

AbbVie has announced a collaboration with the Netherlands’ Utrecht University and Erasmus Medical Center and the Chinese-Dutch biotech Harbour Biomed to develop a neutralizing antibody that can treat Covid-19. The antibody, called 47D11, was discovered by AbbVie’s three partners, and AbbVie will support early preclinical work, while preparing for later preclinical and clinical development. Researchers described the antibody in Nature Communications last month.

Before remdesivir soaked up the spotlight amid the coronavirus crisis, Gilead’s filgotinib was the star experimental drug tapped to rake in billions competing with other JAK inhibitors made by rivals including AbbVie and Eli Lilly.

Now, long term data on the drug — discovered by Gilead’s partners at Galapagos and posted as part of a virtual medical conference — have solidified the durability and safety of filgotinib in patients with rheumatoid arthritis, spanning data from three late-stage trials. An FDA decision on the drug is expected this year.

→ Merck — one of the last big pharma bastions in the beleaguered field of antibiotic drug development — on Friday said the FDA had signed off on using its combination drug, Recarbrio, with hospital-acquired bacterial pneumonia and ventilator-associated bacterial pneumonia. The drug could come handy for use in hospitalized patients who are afflicted with Covid-19, who carry a higher risk of contracting secondary bacterial infections. Once SARS-CoV-2, the virus behind Covid-19, infects the airways, it engages the immune system, giving other pathogens free rein to pillage and plunder as they please — the issue is particularly pertinent in patients on ventilators, which in any case are breeding grounds for infectious bacteria.