23andMe co-founder’s new startup, Precise.ly, brings genomics to India through Narayana partnership

Precise.ly, the new genomics startup launched by 23andMe co-founder Linda Avey and Aneil Mallavarapu, is taking its spin on direct to consumer personalized genomics to India through a partnership with Naryana Health, one of India’s leading specialty hospital networks.

Narayana, a company that operates a network of 24 hospitals serving 2.5 million patients, is one of the most fascinating stories in healthcare. By emphasizing efficiencies and cost savings, the hospital network has managed to bring costs down dramatically for many procedures — including providing cancer surgeries for as little as $700 and heart bypass surgeries for $3,000 (as this fascinating article in Bloomberg BusinessWeek illustrates).

Precise.ly’s mission — to collect and analyze genetic data from populations that typically haven’t had access to the services — is one that resonates in a world where the majority of research has been conducted on wealthier populations in wealthy countries. Other startups, like 54Gene, are trying to bring a similar message to the African continent.

“To date, most human genetics research has focused on European populations. But genetic insights need to be tuned to the rest of the world,” said Mallavarapu, in a statement. “We’ve assembled a team of experts who are pioneering advances in genetic analysis and its application to the huge populations of people in south Asia and beyond.” 

Some of that work is being done in concert with Narayana health, the hospital network founded by Dr. Devi Prasad Shetty nearly twenty years ago. Dr. Shetty is initially hoping that Precise.ly’s genetic database will be able to help his hospitals build out a stem cell donor registry that could help hundreds of thousands of Indians who need transplants.

“Personal genetic testing is recognized by the U.S. FDA to test genetic risk for Parkinsonism, late onset Alzheimer’s disease and celiac disease. It is only a matter of time before most diseases get added to the list,” Dr. Shetty said in a statement. “Because of the simplicity of genetic testing from saliva samples, it’s possible to conduct large-scale population screening at a reasonable cost. We are working with Precise.ly’s team of researchers to add HLA typing, which has the potential to transform cancer and other disease treatments in India.”

The path to entering the Indian market was slightly circuitous for Precise.ly. When Avey first left 23andMe, she went to RockHealth (an investor in the company’s $1 million seed round), and began exploring ways to organize and store more of a patient’s quantified health data.

As that company failed to gain traction, Avey took another look at the genetics market and found that there were significant opportunities in underserved markets — and that India, with its rising middle class and burgeoning healthcare industry would be a good target.

“We decided we would build on this Helix platform all kinds of apps for people who had specific diagnosis,” says Avey. But the market was already chock full of startups (including 23andMe), so an early investor in the company from, Civilization Ventures, and its founder Shahram Seyedin-Noor suggested that they begin to look globally for growth.

“Precise.ly’s mission is to deliver validated genetic insights to the billions of people living outside the western world. We’re initially focused on India where there are urgent health issues readily addressable through access to personal genomic data,” said Avey, the chief executive officer of Precise.ly, in a statement. “Our partnership with Narayana is vital to delivering on the promise of precise, data-driven health.” 

Xilis believes cultivating micro-tumors may hold the key to more effective cancer treatments

Despite near-miraculous advances in the treatment of cancer in the U.S. and around the world, the disease remains the second leading cause of death in America.

The problem is that every manifestation of the disease is unique to the patient that is afflicted by it, because everyone’s body is actually different. Most cancer treatments are determined by their ability to cure the largest population afflicted with a particular cancer type.

As understanding of the disease has advanced more targeted treatments are coming to market to treat particular types of the disease. And now, Xilis has developed a process that its founders hope will make those treatments even more effective.

Founded by Xiling Shen and Dr. David Hsu, two Duke University professors and researchers, the company’s technology is based on research conducted by the Dutch research scientist Hans Clevers. Clevers, who won the Breakthrough Prize for life sciences in 2004 and serves on the board of directors of Roche, helped refine a technique for growing small versions of human organs for research.

Shen and Hsu have taken that research and advanced it, developing a process which can cultivate and sustain tumors from a cancer patient — allowing physicians and pharmaceutical companies to develop even more tailored treatments that can respond to the particular type of cancer.

“Our technology creates 10,000 micro tumors from a single cancer biopsy which then tests which cancer treatments will or won’t work for a patient, saving them critical time in his/her cancer treatment plan,” said Shen in a statement. “Already in clinical trials, we have data showing our technology can successfully predict treatment success and finding new therapy for drug resistant patients.”

Shen and his co-founder first initiated clinical trials on their new discovery early in 2019. Their results were so promising that the two decided to form a company around the innovation and raise capital to accelerate the time-to-clinic so that patients could reap the benefits of these more targeted therapies.

Indeed, the technology is so compelling that Clevers, the progenitor of the company’s technology has agreed to join the company as a co-founder and collaborate on future development, according to an interview with Shen.

“What we have invented is this microfluidics droplet,” says Shen. “We’re growing miniature organoids so these cancer cells are growing in a 3D tumor micro environment.”

To accelerate the commercialization of the technology, Xilis has raised $3 million in seed funding from investors including Felicis Ventures, an early investor in the multi-billion dollar cancer treatment technology developer Guardant Health, former NFL superstar Joe Montana’s Liquid 2 Ventures fund, along with Pear and 8VC. As a result, Chan will take a seat on the company’s board of directors.

While the near-term value of the company’s technology is in its ability to better target therapies for patients, longer term, there’s value in the data set that the company is amassing. “We’re accumulating a micro-organoid bank that pharma would love to test these things on,” says Shen.

The potential to save pharmaceutical companies millions of dollars to do initial testing on how effective a treatment is can’t be overlooked, according to Shen. Using the technology, pharma companies “can do massive drug screening at a much higher throughput with a much lower cost.” 


Paige raises $45M more to map the pathology of cancer using AI

One of the more notable startups using artificial intelligence to understand and fight cancer has raised $45 million more in funding to continue building out its operations and inch closer to commercialising its work.

Paige — which applies AI-based methods such as machine learning to better map the pathology of cancer, an essential component of understanding the origins and progress of a disease with seemingly infinite mutations (its name is an acronym of Pathology AI Guidance Engine) — says it will be use the funding to inch closer to FDA approvals for products it is developing in areas such as biomarkers and prognostic capabilities.

It also plans to use the funding to continue developing better ways of diagnosing and ultimately fighting the disease, as well as exploring further commercial opportunities for its work, specifically within the bio-pharmaceutical industry.

This round is being led by Healthcare Venture Partners, with previous investor Breyer Capital, Kenan Turnacioglu and other funds participating. The company is not disclosing its valuation, but PitchBook noted that a first close of this round (when it raised $33 million) put the valuation at $208 million. That would value Paige now at about $220 million with the $45 million close, more than three times its valuation in its previous round.

Paige first emerged from stealth back in 2018 — with a bang.

Paige.AI — as it was known at the time — was hatched inside the Memorial Sloan Kettering Cancer Center, one of the world’s foremost institutions both for working on cancer therapies and treating cancer patients, and along with a $25 million investment led by Jim Breyer, Paige had secured exclusive access to MSK’s 25 million pathology slides as well as its intellectual property related to the AI-based computational pathology that underpinned its work. These slides make up one of the biggest repositories of its kind in the world, and as all solutions and services built on machine learning are only as good as the data that’s fed into them, they were critical to the startup’s beginnings.

The startup also launched with some serious talent behind it.

Much of the computational pathology being used by Paige had been developed by Dr Thomas Fuchs, who is known as the “father of computational pathology” and is the director of Computational Pathology in The Warren Alpert Center for Digital and Computational Pathology at Memorial Sloan Kettering, as well as a professor of machine learning at the Weill Cornell Graduate School of Medical Sciences.

Fuchs co-founded Paige with Dr David Klimstra, chairman of the department of pathology at MSK, and Fuchs had originally started out as the CEO of Paige, but was replaced earlier this year by Leo Grady, who joined from another bio-startup, Heartflow (another company backed by Healthcare Venture Partners). Fuchs is still supporting the company, but no longer in an executive role.

In the nearly two years since it launched, there have been some milestones reached. The company, which has around 30 employees today, has been the first to get an FDA breakthrough designation (which helps expedite the long process of drug approvals in urgent areas where there are few or no other options for patients) for using AI in oncology pathology. It’s also the first to get a CE mark in the same category, which opens the door to working in Europe, too. Paige has so far ingested 1.2 million images into its slide database and is using them — in algorithms that also take in genomic data, drug response data and outcome data — to work on developing diagnostic solutions.

But as with all new medical products, progress is not measured in quarters as it might be with a more typical tech startup. Moving fast and breaking things is something to be avoided. So even with all of the above advances, there has yet to be any commercial products launched, nor is Grady giving any specific time frames for when they will. And when the company came out of stealth in 2018, it said it would be focusing on breast, prostate and other major cancers, although today it’s not as quick to specify what its targets will be when it does launch commercial products.

Similarly, it’s also expanding its remit from primarily clinical environments to pharmaceutical ones.

“The clinical side is still our focus, but this is an expansion and realisation that this has a broader impact, and that includes pharmaceutical customers,” Grady said. 

And the dropping of the .AI in its name was also intentional, in part a reaction it seems to how much AI gets thrown around today.

“There is a fundamental misconception, which is thinking of AI as a product and not a technology,” said Grady. “It’s a technology set that can allow you to do many things that could not have been done in the past, but you need to apply it in a meaningful way. Developing a good AI and putting that on the market will not cut it in terms of clinical adoption.”

The funding round, Grady said, saw a lot of interest from strategic investors, although the company intentionally has stayed away from these.

“We were approached by all of the scanner vendors and some of the biopharmaceutical companies,” he said. “But we made the decision to not take a strategic investment with this round because we wanted to be neutral with hardware vendors and not be too tied with any one.”

He also pointed to the challenges of talking to investors when you are working in a cutting-edge area (a challenge that has foxed many an investor also into backing the wrong horses, too, such as Theranos).

“We’re at the intersection of three areas: tech, medical devices and clinical medicine, and life sciences and biotech,” he said. “Many investors sit squarely in one and don’t feel comfortable in others. That makes the conversations challenging and short. But there has been an increasing blend between those three sectors.”

That’s where Healthcare Venture Partners fits into the mix. “Paige exemplifies the benefits of digital pathology and represents the bright future of AI-driven medical diagnosis,” said Jeff Lightcap of Healthcare Venture Partners, in a statement. “As hospitals embark on digital transformations, they will face challenges associated with these transitions. We believe Paige addresses many of these issues by enhancing the ability of clinical teams and pathologists to collaborate. We’re confident in Paige’s future and believe they will continue to develop cutting-edge technologies that enable pathology departments to transform their practices, which have changed little in the last century.”

“We applaud Paige’s commitment to building clinical AI products that will improve the diagnostic process and patient care,” added Jim Breyer of Breyer Capital, in a statement. “This is a critical time for Pathology, as pathologists are carrying a heavier workload than ever before. Paige understands their needs and the team has built cutting-edge technologies to address them. Paige represents the future of computational pathology and we look forward to their continued growth and success.”

Trialjectory uses self-reported clinical data to match cancer patients with clinical trials

TrialJectory, which is developing a new technology service to match cancer patients with clinical trials, has raised $2.7 million to finance its continued growth.

Led by Contour Venture Partners, the new financing will be used to accelerate TrialJectory’s operations by adding more clinical trials for different cancer types and expanding the company’s outreach to caregivers, pharmaceutical companies and patients, the company said.

“As cancer is the second leading cause of death for Americans – with thousands of new cases diagnosed each year – having access to advanced treatment options is a necessity, not a privilege, as new trials provide better outcomes to patients,” said Tzvia Bader, TrialJectory chief executive and co-founder. “What’s more, one of the top obstacles that oncologists face today is the lack of clinical trial access for patients, which is due to the availability of more treatment options overall. Additionally, it is a very complex process to match the right patient with the right treatment, especially with the rise of personalized medicine.”

The company currently supports trials for breast cancer, colon cancer, bladder cancer, melanoma, and myelodysplastic syndromes.

TrialJectory’s software was trained to seek out keywords in unstructured treatment descriptions and extracting relevant data. Its software then groups that information into clusters and standardizes the information to create a database that highlights patient attributes that would be appropriate for clinical trials.

Patients are then matched to the clinical trials after filling out a questionnaire.

“TrialJectory’s work – driven by a highly experienced management team, comprised of both oncology and technology experts – is disrupting and reshaping how we think about traditional cancer care today,” concluded Bob Greene, from Contour Venture Partners. “Even more important, it is empowering patients to take back control of their treatment, and we look forward to watching TrialJectory’s platform continue to grow quickly. We believe that the Company has the potential to become a go-to resource for the global medical community to help doctors provide personalized, matched treatment options to patients in need everywhere.”

You’ve heard of CRISPR, now meet its newer, savvier cousin CRISPR Prime

CRISPR, the revolutionary ability to snip out and alter genes with scissor-like precision, has exploded in popularity over the last few years and is generally seen as the standalone wizard of modern gene-editing. However, it’s not a perfect system, sometimes cutting at the wrong place, not working as intended and leaving scientists scratching their heads. Well, now there’s a new, more exacting upgrade to CRISPR called Prime, with the ability to, in theory, snip out more than 90% of all genetic diseases.

Just what is this new method and how does it work? We turned to IEEE fellow, biomedical researcher and dean of graduate education at Tuft University’s school of engineering Karen Panetta for an explanation.

How does CRISPR Prime editing work?

CRISPR is a powerful genome editor. It utilizes an enzyme called Cas9 that uses an RNA molecule as a guide to navigate to its target DNA. It then edits or modifies the DNA, which can deactivate genes or insert a desired sequence to achieve a behavior. Currently, we are most familiar with the application of genetically modified crops that are resistant to disease.

However, its most promising application is to genetically modify cells to overcome genetic defects or its potential to conquer diseases like cancer.

Some applications of genome editing technology include:

  • Genetically modified mosquitos that can’t carry malaria.
  • In humans, “turning on” a gene that can create fetal type behaving cells that can overcome sickle-cell anemia.

Of course, as with every technology, CRISPR isn’t perfect. It works by cutting the double-stranded DNA at precise locations in the genome. When the cell’s natural repair process takes over, it can cause damage or, in the case where the modified DNA is inserted at the cut site, it can create unwanted off-target mutations.

Some genetic disorders are known to mutate specific DNA bases, so having the ability to edit these bases would be enormously beneficial in terms of overcoming many genetic disorders. However, CRISPR is not well suited for intentionally introducing specific DNA bases, the As, Cs, Ts and Gs that make up the double helix.

Prime editing was intended to overcome this disadvantage, as well as other limitations of CRISPR.

Prime editing can do multi-letter base-editing, which could tackle fatal genetic disorders such as Tay-Sachs, which is caused by a mutation of four DNA letters.

It’s also more precise. I view this as analogous to the precision lasers brought to surgery versus using a hand-held scalpel. It minimized damage, so the healing process was more efficient.

Prime editing can insert, modify or delete individual DNA letters; it also can insert a sequence of multiple letters into a genome with minimal damage to DNA strands.

How effective might Prime editing be?

Imagine being able to prevent cancer and/or hereditary diseases, like breast cancer, from ever occurring by editing out the genes that are makers for cancer. Cancer treatments are usually long, debilitating processes that physically and emotionally drain patients. It also devastates patients’ loved ones who must endure watching helpless on the sidelines as the patient battles to survive.

“Editing out” genetic disorders and/or hereditary diseases to prevent them from ever coming to fruition could also have an enormous impact on reducing the costs of healthcare, effectively helping redefine methods of medical treatment.

It could change lives so that long-term disability care for diseases like Alzheimer’s and special needs education costs could be significantly reduced or never needed.

How did the scientific community get to this point — where did CRISPR/prime editing “come from?”

Scientists recognized CRISPR’s ability to prevent bacteria from infecting more cells and the natural repair mechanism that it initiates after damage occurs, thus having the capacity to halt bacterial infections via genome editing. Essentially, it showed adaptive immunity capabilities.

When might we see CRISPR Prime editing “out in the wild?”

It’s already out there! It has been used for treating sickle-cell anemia and in human embryos to prevent HIV infections from being transmitted to offspring of HIV parents.

So, what’s next?

IEEE engineers, like myself, are always seeking to take the fundamental science and expand it beyond the petri dish to benefit humanity.

In the short term, I think that Prime editing will help generate the type of fetal like cells that are needed to help patients recover and heal as well as developing new vaccines against deadly diseases. It will also allow researchers new, lower cost alternatives and access to Alzheimer’s like cells without obtaining them post-mortem.

Also, AI and deep learning is modeled after human neural networks, so the process of genome editing could potentially help inform and influence new computer algorithms for self-diagnosis and repair, which will become an important aspect of future autonomous systems.

Facebook unveils its first foray into personal digital healthcare tools

Nearly a year and a half after the Cambridge Analytica scandal reportedly scuttled Facebook’s fledgling attempts to enter the healthcare market, the social media giant is launching a tool called “Preventive Health” to prompt its users to get regular checkups and connect them to service providers.

The architect of the new service is Dr. Freddy Abnousi, the head of the company’s healthcare research, who was previously linked to an earlier skunkworks initiative that would collect anonymized hospital data and use a technique called “hashing” to match the data to individuals that exist in both data sets — for research, according to CNBC reporting.

Working with the American Cancer Society; the American College of Cardiology; the American Heart Association; and the Centers for Disease Control and Prevention Facebook is developing a series of digital prompts that will encourage users to get a standard battery of tests that’s important to ensure health for populations of a certain age.

The company’s initial focus is on the top two leading causes of death in the U.S.: heart disease and cancer — along with the flu, which affects millions of Americans each year.

“Heart disease is the number one killer of men and women around the world and in many cases it is 100% preventable. By incorporating prevention reminders into platforms people are accessing every day, we’re giving people the tools they need to be proactive about their heart health,” said Dr. Richard Kovacs, the president of the American College of Cardiology, in a statement.

Users who want to access Facebook’s Preventive Health tools can search in the company’s mobile app to find which checkups are recommended by the company’s partner organizations based on the age and gender of a user.

The tool allows Facebookers to mark when the tests are completed, set reminders to schedule future tests and tell people in their social network about the tool.

Facebook will even direct users to resources on where to have the tests. One thing that the company will not do, Facebook assures potential users, is collect the results of any test.

“Health is particularly personal, so we took privacy and safety into account from the beginning. For example, Preventive Health allows you to set reminders for your future checkups and mark them as done, but it doesn’t provide us, or the health organizations we’re working with, access to your actual test results,” the company wrote in a statement. “Personal information about your activity in Preventive Health is not shared with third parties, such as health organizations or insurance companies, so it can’t be used for purposes like insurance eligibility.”

The company said that people can also use the new health tool to find locations that administer flu shots.

“Flu vaccines can have wide-ranging benefits beyond just preventing the disease, such as reducing the risk of hospitalization, preventing serious medical events for some people with chronic diseases, and protecting women during and after pregnancy,” said Dr. Nancy Messonnier, Director, National Center for Immunization and Respiratory Diseases, CDC, in a statement. “New tools like this will empower users with instant access to information and resources they need to become a flu fighter in their own communities.”

The UK’s National Health Service is launching an AI lab

The UK government has announced it’s rerouting £250M (~$300M) in public funds for the country’s National Health Service (NHS) to set up an artificial intelligence lab that will work to expand the use of AI technologies within the service.

The Lab, which will sit within a new NHS unit tasked with overseeing the digitisation of the health and care system (aka: NHSX), will act as an interface for academic and industry experts, including potentially startups, encouraging research and collaboration with NHS entities (and data) — to drive health-related AI innovation and the uptake of AI-driven healthcare within the NHS. 

Last fall the then new in post health secretary, Matt Hancock, set out a tech-first vision of future healthcare provision — saying he wanted to transform NHS IT so it can accommodate “healthtech” to support “preventative, predictive and personalised care”.

In a press release announcing the AI lab, the Department of Health and Social Care suggested it would seek to tackle “some of the biggest challenges in health and care, including earlier cancer detection, new dementia treatments and more personalised care”.

Other suggested areas of focus include:

  • improving cancer screening by speeding up the results of tests, including mammograms, brain scans, eye scans and heart monitoring
  • using predictive models to better estimate future needs of beds, drugs, devices or surgeries
  • identifying which patients could be more easily treated in the community, reducing the pressure on the NHS and helping patients receive treatment closer to home
  • identifying patients most at risk of diseases such as heart disease or dementia, allowing for earlier diagnosis and cheaper, more focused, personalised prevention
  • building systems to detect people at risk of post-operative complications, infections or requiring follow-up from clinicians, improving patient safety and reducing readmission rates
  • upskilling the NHS workforce so they can use AI systems for day-to-day tasks
  • inspecting algorithms already used by the NHS to increase the standards of AI safety, making systems fairer, more robust and ensuring patient confidentiality is protected
  • automating routine admin tasks to free up clinicians so more time can be spent with patients

Google-owned UK AI specialist DeepMind has been an early mover in some of these areas — inking a partnership with a London-based NHS trust in 2015 to develop a clinical task management app called Streams that’s been rolled out to a number of NHS hospitals.

UK startup, Babylon Health, is another early mover in AI and app-based healthcare, developing a chatbot-style app for triaging primary care which it sells to the NHS. (Hancock himself is a user.)

In the case of DeepMind, the company also hoped to use the same cache of NHS data it obtained for Streams to develop an AI algorithm for earlier detection of a condition called acute kidney injury (AKI).

However the data-sharing partnership ran into trouble when concerns were raised about the legal basis for reusing patient data to develop AI. And in 2017 the UK’s data watchdog found DeepMind’s partner NHS trust had failed to obtain proper consents for the use of patients’ data.

DeepMind subsequently announced its own AI model for predicting AKI — trained on heavily skewed US patient data. It has also inked some AI research partnerships involving NHS patient data — such as this one with Moorfields Eye Hospital, aiming to build AIs to speed up predictions of degenerative eye conditions.

But an independent panel of reviewers engaged to interrogate DeepMind’s health app business raised early concerns about monopoly risks attached to NHS contracts that lock trusts to using its infrastructure for delivering digital healthcare.

Where healthcare AIs are concerned, representative clinical data is the real goldmine — and it’s the NHS that owns that.

So, provided NHSX properly manages the delivery infrastructure for future digital healthcare — to ensure systems adhere to open standards, and no single platform giant is allowed to lock others out — Hancock’s plan to open up NHS IT to the next wave of health-tech could deliver a transformative and healthy market for AI innovative that benefits startups and patients alike.

Commenting on the launch of NHSX in a statement, Hancock said: “We are on the cusp of a huge health tech revolution that could transform patient experience by making the NHS a truly predictive, preventive and personalised health and care service.

“I am determined to bring the benefits of technology to patients and staff, so the impact of our NHS Long Term Plan and this immediate, multimillion pound cash injection are felt by all. It’s part of our mission to make the NHS the best it can be.

“The experts tell us that because of our NHS and our tech talent, the UK could be the world leader in these advances in healthcare, so I’m determined to give the NHS the chance to be the world leader in saving lives through artificial intelligence and genomics.”

Simon Stevens, CEO of NHS England, added: “Carefully targeted AI is now ready for practical application in health services, and the investment announced today is another step in the right direction to help the NHS become a world leader in using these important technologies.

“In the first instance it should help personalise NHS screening and treatments for cancer, eye disease and a range of other conditions, as well as freeing up staff time, and our new NHS AI Lab will ensure the benefits of NHS data and innovation are fully harnessed for patients in this country.”

Google’s SMILY is reverse image search for cancer diagnosis

Spotting and diagnosing cancer is a complex and difficult process even for the dedicated medical professionals who do it for a living. A new tool from Google researchers could improve the process by providing what amounts to reverse image search for suspicious or known cancerous cells. But it’s more than a simple matching algorithm.

Part of the diagnosis process is often examining tissue samples under a microscope and looking for certain telltale signals or shapes that may indicate one or another form of cancer. This can be a long and arduous process because every cancer and every body is different, and the person inspecting the data must not only look at the patient’s cells but also compare them to known cancerous tissues from a database or even a printed book of samples.

As has been amply demonstrated for years now, matching similar images to one another is a job well suited to machine learning agents. It’s what powers things like Google’s reverse image search, where you put in one picture and it finds ones that are visually similar. But this technique has also been used to automate processes in medicine, where a computer system can highlight areas of an X-ray or MRI that have patterns or features it has been trained to recognize.

That’s all well and good, but the complexity of cancer pathology rules out simple pattern recognition between two samples. One may be from the pancreas, another from the lung, for example, meaning the two situations might be completely different despite being visually similar. And an experienced doctor’s “intuition” is not to be replaced, nor would the doctor suffer it to be replaced.

Aware of both the opportunities and limitations here, Google’s research team built SMILY (Similar Medical Images Like Yours), which is a sort of heavily augmented reverse image search built specifically for tissue inspection and cancer diagnosis.

A user puts into the system a new sample from a patient — a huge, high-resolution image of a slide on which a dyed section of tissue is laid out. (This method is standardized and has been for a long time — otherwise how could you compare any two?)


Once it’s in the tool, the doctor can inspect it as they would normally, zooming in and panning around. When they see a section that piques their interest, they can draw a box around it and SMILY will perform its image-matching magic, comparing what’s inside the box to the entire corpus of the Cancer Genome Atlas, a huge database of tagged and anonymized samples.

Similar-looking regions pop up in the sidebar, and the user can easily peruse them. That’s useful enough right there. But as the researchers found out while they were building SMILY, what doctors really needed was to be able to get far more granular in what they were looking for. Overall visual similarity isn’t the only thing that matters; specific features within the square may be what the user is looking for, or certain proportions or types of cells.

As the researchers write:

Users needed the ability to guide and refine the search results on a case-by-case basis in order to actually find what they were looking for…This need for iterative search refinement was rooted in how doctors often perform “iterative diagnosis”—by generating hypotheses, collecting data to test these hypotheses, exploring alternative hypotheses, and revisiting or retesting previous hypotheses in an iterative fashion. It became clear that, for SMILY to meet real user needs, it would need to support a different approach to user interaction.

To this end the team added extra tools that let the user specify much more closely what they are interested in, and therefore what type of results the system should return.

First, a user can select a single shape within the area they are concerned with, and the system will focus only on that, ignoring other features that may only be distractions.

Second, the user can select from among the search results one that seems promising and the system will return more like it, less closely tied to the original query. This lets the user go down a sort of rabbit hole of cell features and types, doing that “iterative” process the researchers mentioned above.


And third, the system was trained to understand when certain features are present in the search result, such as fused glands, tumor precursors, and so on. These can be included or excluded in the search — so if someone is sure it’s not related to this or that feature, they can just sweep all those examples off the table.

In a study of pathologists given the tool to use, the results were promising. The doctors appeared to adopt the tool quickly, not only using its official capabilities but doing things like reshaping the query box to test the results or see if their intuition on a feature being common or troubling was right. “The tools were preferred over a traditional interface, without a loss in diagnostic accuracy,” the researchers write in their paper.

It’s a good start, but clearly still only an experiment. The processes used for diagnosis are carefully guarded and vetted; you can’t just bring in a random new tool and change up the whole thing when people’s lives are on the line. Rather, this is merely a bright start for “future human-ML collaborative systems for expert decision-making,” which may at some point be put into service at hospitals and research centers.

You can read the two papers describing SMILY and the doctor-focused refinements to SMILY here; they were originally presented at CHI 2019 in Glasgow earlier this year.

Paige details first AI pathology tech with clinical-grade accuracy in new research paper

Medical tech and computational pathology startup Paige has published a new article in the peer-reviewed medical journal Nature Medicine, detailing its artificial intelligence-based detection system for identifying prostate cancer, skin cancer and breast cancer, which the company says achieves “near-perfect accuracy.” Paige’s tech, which employs deep learning trained on a dataset of almost 45,000 slide images taken from over 15,000 patients spanning 44 countries, is novel in that it can eschew the need to curate data sets for training first, which greatly decreases cost and time required to build accurate AI-based diagnostic tools.

Last February, Paige announced $25 million in Series A funding, and a partnership with Memorial Sloan Kettering Center (MSK) to gain access to one of the largest single repositories of pathology slides in the world. MSK is also home to the lab of Dr. Thomas Fuchs, Paige’s co-founder and Chief Scientific Officer, and possibly the world’s foremost authority in computational pathology.

Paige’s approach uses much larger data sets than are typically employed in AI-based diagnostics, but without the tight curation that focuses other efforts much more narrowly on specific types of cancer diagnostics. The result, according to the company, is not only better performance, but also a resulting system that its much more generally applicable.

Next up for Paige is to commercialize its technology, which is something it’s already pursuing. The work described in the article published in Nature Medicine has already been employed in technology currently under review by the FDA, albeit for a different final application than the ones described in the study published by the magazine.


Careteam aims to unite patients and healthcare providers with a platform approach

How best to untangle the Gordian knot that is navigating your own healthcare? It’s a tricky question, and one that seems to have become only more complicated as technology improves, in many regards — systems don’t necessarily speak to one another, and it’s still hard for an ordinary patient without specialist knowledge to make sense of everything. Careteam is a Canadian startup hoping to address that, looking to replicate the kind of advances made possible by technology in industries like e-commerce and enterprise software.

Careteam co-founder and CEO Dr. Alexandra Greenhill has experienced the frustration of being a tech-savvy person in a world of healthcare that can seem technologically inept — both as a practicing GP and as someone who depends on the healthcare system as a patient and a relative of patients with more sophisticated medical needs.

“I spent more than 15 years innovating within the healthcare system,” Greenhill told me in an interview. “I computerized hospitals, helped doctors adopt electronic medical records and other types of innovation practices. And then for the last eight years, I’ve been in tech, trying to figure out how to build the kind of technology we need in health, and especially digital health.”

All that experience led Greenhill to the realization that while there were many companies building specific solutions for real, but relatively narrow problems, that didn’t reflect how most people experienced care. Greenhill and her team of three other co-founders (Jeremy P. Smith, Robert I. Atwell and Kevin Lysyk) had all had unfortunate, but eye-opening experiences with family members in need of treatment for major diseases.

“You step in and you discover that cancer care, palliative care, post-surgical care — there’s so many things that would have gone wrong if we didn’t have the expertise ourselves,” Greenhill said. “But in the meantime, you end up being sort of pulled into multiple directions and saying ‘this makes no sense.’ You know, I can purchase stuff online in my private life; I can use all kinds of tools in the business world, and yet it’s back to paper and voice in health, which matters most.”

Careteam CEO and founder Dr. Alexandra Greenhill

What Careteam provides is collaboration for care — true collaboration, designed to span patients, their doctors and other healthcare pros, their families and anyone who matters to them in the course of pursuing their care. It provides the ability to communicate instantly, build care plans that integrate all aspects of their tailored health plans, receive custom-configurable notifications and measure progress toward specific goals set by patient and healthcare providers.

Part of the reason this process has become opaque or difficult is precisely due to innovation: Greenhill takes issue with the prevailing narrative that the healthcare industry is somehow allergic to innovation.

“There’s this sort of perception that healthcare doesn’t innovate, but it’s also almost insulting to the healthcare system, because we have innovated — we save people from cancer, where we couldn’t,” she noted. “We cure HIV, in some cases, and we prevent it from being transmitted to unborn babies of mothers with full-blown AIDS and things that in my working lifetime were impossibilities; it was science fiction to help someone with HIV. And, and we’ve managed to do all of that, and it’s a success story. We’ve created complexity, we’ve created people who live with 12 conditions for many, many years and take complicated drug regiments.”

In addition to advances in treatment, Greenhill notes that she and her team couldn’t have build Careteam five years ago, because cloud storage wasn’t secure and everything had to be done on a site-specific instance, and that would’ve been cost-prohibitive to build. In other words, technology has been applied to, and vastly improved, healthcare overall, regardless of the general perception of the industry as an innovation laggard.

That’s why Greenhill’s startup doesn’t shy away from complexity — they embrace it. Careteam is designed not to try to normalize and standardize the varied and highly specialized landscape of healthcare solutions and providers through anything like a one-size-fits-all API. Instead, the company’s tech development is cleverly designed to be flexible when it comes to integrations.

“We collectively spent $1.9 billion in Canada, to try and digitize the healthcare system, create standards and create some exchange between data,” Greenhill said. “The NHS tried the same, big U.S. hospital systems have created their own little sort of islands, including Kaiser and Mayo and others. And the conclusion of all of that is standardization in healthcare just doesn’t seem to catch on.”

Careteam’s approach has been instead to integrate specific clinics, and let practitioners and patients derive benefits and help spur the adoption of the platform to their companion organizations and clinics. It’s a sort of rhizomatic approach that starts with a node central to a patient’s care and spreads through the healthcare professionals and members of the patient’s support network that the product helps. And integration is made possible without technical demands on the part of partners thanks to the work of CTO Lysyk, according to Greenhill.

The Vancouver-based startup is working with the Centre for Aging + Brain Health in Toronto, Ontario in a validation program announced last year, and also raised an initial round of funding in January led by BCF Ventures with participation from Right Side Capital, Globalive Capital, Atrium Ventures, and angels Barney Pell and Ajay Agarwal .