Thanks to the sheer amount of data that machine learning technologies collect, end-user privacy will be more important than ever.

It's still very early days for artificial intelligence (AI) in businesses. But the data that desktop and mobile applications automatically collect, analyze using machine learning algorithms and act upon is a reality, and IT shops must be ready to handle this type and volume of information. In particular, thorny artificial intelligence data privacy issues can arise if employers can detect and view more -- and more personal -- data about their employees on devices or apps.

"AI requires a ton of data, so the privacy implications are bigger," said Andras Cser, vice president and principal analyst at Forrester Research. "There's potential for a lot more personally identifiable data being collected. IT definitely needs to pay attention to masking that data."

Business applications and devices can take advantage of machine learning in a number of ways. A mobile sales app could collect location or IP address data and find patterns to connect the user with customers in their area, for instance. If the user accesses this app on a personal device they use for work, they may not want their employer to be able to view that data when they're off the clock. Or, a user's personal apps could learn information about the individual that he or she wouldn't want their human resources department to find out.

Health-related devices that take advantage of artificial intelligence pose a significant threat. A lot of companies give out Fitbits, for example, to gather data about employees that's used for insurance purposes, said mobile expert Brian Katz. Artificial intelligence data from that kind of device could reveal a health condition the employer didn't know about, and then comes the real dilemma:

"If your manager knows about it, do they act on it?" Katz said.

One way for IT to address data privacy issues with machine learning is to "mask" the data collected, or anonymize it so that observers can't learn specific information about a specific user. Some companies take a similar approach now with regulatory compliance, where blind enforcement policies use threat detection to determine if a device follows regulations but do not glean any identifying information.

With AI it becomes easier to correlate data ... and remove privacy. Brian Katzmobile expert

Device manufacturers have also sought to protect users in this way. For example, Apple iOS 10 added differential privacy, which recognizes app and data usage patterns among groups of users while obscuring the identities of individuals.

"If you know a couple things that you can correlate, you can identify a person," Katz said. "With AI it becomes easier to correlate data ... and remove privacy. People want to provide a better experience and learn more about [users], and doing that in an anonymous way is very difficult."

Tools such as encryption are also important for IT to maintain data privacy and security, Cser said. IT departments should also have policies in place that make it clear to users what is permissible and not permissible data for IT to collect and what the business can do with it, he said.

It's important for users to understand this information, Katz said.

"Part of it's just being transparent with users about what you're doing with the data," he said.

Another best practice is to separate business and personal apps using technologies such as containerization, he said. Enterprise mobility management tools can be set up to look at only corporate apps but still be able to whitelist and blacklist any apps to prevent malware. That way, IT doesn't invade users' privacy on personal apps.

Privacy regulations vary widely across the globe, and many businesses and countries are still working to update guidelines based on emerging technology.

The European Union (EU), for example, has strong protection for the personal privacy of employees. Individuals must be notified of any data gathered about them, any data processing can be done only if there is a "legitimate" purpose such as investigating suspected criminal activity, and collected data must be kept secure. There are also restrictions on entities sharing collected data outside the EU.

The United States is more lax, said Joseph Jerome, a policy counsel on the Privacy & Data Project at the Center for Democracy & Technology in Washington, D.C.

"Basically employers can get away with anything they want so long as they're providing some kind of notice of consent," he said.

That's the reason some companies prefer to provide corporate-owned devices rather than enable BYOD, Katz said.

"You don't have as much of an expectation about privacy there, and that's why they do it," he said. "Your privacy is much more limited on a [corporate] device."

And when it comes to artificial intelligence data specifically, an interesting question arises: Who is responsible for the learned information? The employer? The machine learning application itself? The person that created the algorithm? These factors are still up in the air, Cser said.

"Legal frameworks are not yet capable of handling this kind of autonomous information," he said. "It's going to be a precedence-based type of evolution."

Still, the data privacy issues raised by artificial intelligence are not entirely new. The internet of things and big data have been able to glean similarly personal and large volumes of data for years.

"It's basically a continuation of those trends," Jerome said. "It's lots and lots of data being gleaned from a lot of different sources. There's a lot of hype here, but at the end of the day ... I don't know if it raises any new issues."

Rather, machine learning might be a unique way to actually help users manage their data privacy, Jerome said. Privacy assistant apps could allow users to create policies that predict and make inferences over time to decide how and when the user would like their data to be collected and used, or not, according to Carnegie Mellon University research.

"AI might be an amazing way to do privacy management," Jerome said.

Original post:
Artificial intelligence data privacy issues on the rise

Albert Hsiao, M.D., and his colleagues at the University of California, San Diego (USCD) health system had been working for 18 months on anartificial intelligence programdesigned to help doctors identify pneumonia on a chest X-ray. When thecoronavirushit the U.S., they decided to see what it could do.

The researchers quickly deployed the application, which dots X-ray images with spots of color where there may be lung damage or other signs of pneumonia. It has now been applied to more than 6,000 chest X-rays, and its providing some value in diagnosis, said Hsiao, director of UCSDs augmented imaging and artificial intelligence data analytics laboratory.

His team is one of several around the country that has pushed AI programs developed in a calmer time into the COVID-19 crisis to perform tasks like deciding which patients face the greatest risk of complications and which can be safely channeled into lower-intensity care.

ASCO Explained: Expert predictions and takeaways from the world's biggest cancer meeting

Join FiercePharma for our ASCO pre- and post-show webinar series. We'll bring together a panel of experts to preview what to watch for at ASCO. Cancer experts will highlight closely watched data sets to be unveiled at the virtual meeting--and discuss how they could change prescribing patterns. Following the meeting, well do a post-show wrap up to break down the biggest data that came out over the weekend, as well as the implications they could have for prescribers, patients and drugmakers.

The machine-learning programs scroll through millions of pieces of data to detect patterns that may be hard for clinicians to discern. Yet few of the algorithms have been rigorously tested against standard procedures. So while they often appear helpful, rolling out the programs in the midst of a pandemic could be confusing to doctors or even dangerous for patients, some AI experts warn.

AI is being used for things that are questionable right now, said Eric Topol, M.D., director of the Scripps Research Translational Institute and author of several books on health IT.

Topol singled out a system created byEpic, a major vendor of electronic health record software, that predicts which coronavirus patients may become critically ill. Using the tool before it has been validated is pandemic exceptionalism, he said.

Epic said the companys model had been validated with data from more 16,000 hospitalized COVID-19 patients in 21 healthcare organizations. No research on the tool has been published, but, in any case, it was developed to help clinicians make treatment decisions and is not a substitute for their judgment, said James Hickman, a software developer on Epics cognitive computing team.

Others see the COVID-19 crisis as an opportunity to learn about the value of AI tools.

My intuition is its a little bit of the good, bad and ugly, said Eric Perakslis, Ph.D., a data science fellow at Duke University and former chief information officer at the FDA. Research in this setting is important.

Nearly $2 billion poured into companies touting advancements in healthcare AI in 2019. Investments in the first quarter of 2020 totaled $635 million, up from $155 million in the first quarter of 2019, according to digital health technology funderRock Health.

At least three healthcare AI technology companies have made funding deals specific to the COVID-19 crisis, includingVida Diagnostics, an AI-powered lung-imaging analysis company, according to Rock Health.

Overall, AIs implementation in everyday clinical care is less common than hype over the technology would suggest. Yet the coronavirus crisis has inspired some hospital systems to accelerate promising applications.

UCSD sped up its AI imaging project, rolling it out in only two weeks.

Hsiaos project, with research funding from Amazon Web Services, the UC system and the National Science Foundation (NSF), runs every chest X-ray taken at its hospital through an AI algorithm. While no data on the implementation have been published yet, doctors report that the tool influences their clinical decision-making about a third of the time, said Christopher Longhurst, M.D., UCSD Healths chief information officer.

The results to date are very encouraging, and were not seeing any unintended consequences, he said. Anecdotally, were feeling like its helpful, not hurtful.

AI has advanced further in imaging than other areas of clinical medicine because radiological images have tons of data for algorithms to process, and more data make the programs more effective, said Longhurst.

But while AI specialists have tried to get AI to do things like predict sepsis and acute respiratory distressresearchers at Johns Hopkins Universityrecently won a NSF grantto use it to predict heart damage in COVID-19 patientsit has been easier to plug it into less risky areas such as hospital logistics.

In New York City, two major hospital systems are using AI-enabled algorithms to help them decide when and how patients should move into another phase of care or be sent home.

AtMount Sinai Health System, an artificial intelligence algorithm pinpoints which patients might be ready to be discharged from the hospital within 72 hours, said Robbie Freeman, vice president of clinical innovation at Mount Sinai.

Freeman described the AIs suggestion as a conversation starter, meant to help assist clinicians working on patient cases decide what to do. AI isnt making the decisions.

NYU Langone Healthhas developed a similar AI model. It predicts whether a COVID-19 patient entering the hospital will suffer adverse events within the next four days, said Yindalon Aphinyanaphongs, M.D., Ph.D., who leads NYU Langones predictive analytics team.

The model will be run in a four- to six-week trial with patients randomized into two groups: one whose doctors will receive the alerts, and another whose doctors will not. The algorithm should help doctors generate a list of things that may predict whether patients are at risk for complications after theyre admitted to the hospital, Aphinyanaphongs said.

Some health systems are leery of rolling out a technology that requires clinical validation in the middle of a pandemic. Others say they didnt need AI to deal with the coronavirus.

Stanford Health Careis not using AI to manage hospitalized patients with COVID-19, saidRon Li, M.D., the centers medical informatics director for AI clinical integration. The San Francisco Bay Areahasnt seen the expected surge of patientswho would have provided the mass of data needed to make sure AI works on a population, he said.

Outside the hospital, AI-enabled risk factor modeling is being used to help health systems track patients who arent infected with the coronavirus but might be susceptible to complications if they contract COVID-19.

At Scripps Health in San Diego, clinicians are stratifying patients to assess their risk of getting COVID-19 and experiencing severe symptoms using a risk-scoring model that considers factors like age, chronic conditions and recent hospital visits. When a patient scores seven or higher, a triage nurse reaches out with information about the coronavirus and may schedule an appointment.

Though emergencies provide unique opportunities to try out advanced tools, its essential for health systems to ensure doctors are comfortable with them and to use the tools cautiously, with extensive testing and validation, Topol said.

When people are in the heat of battle and overstretched, it would be great to have an algorithm to support them, he said. We just have to make sure the algorithm and the AI tool isnt misleading, because lives are at stake here.

Kaiser Health News(KHN) is a national health policy news service. It is an editorially independent program of theHenry J. Kaiser Family Foundationwhich is not affiliated with Kaiser Permanente.

ThisKHNstory first published onCalifornia Healthline, a service of theCalifornia Health Care Foundation

See the article here:
Coronavirus tests the value of artificial intelligence in medicine - FierceBiotech

Dr. Albert Hsiao and his colleagues at the UC San Diego health system had been working for 18 months on an artificial intelligence program designed to help doctors identify pneumonia on a chest X-ray. When the coronavirus hit the United States, they decided to see what it could do.

The researchers quickly deployed their program, which dots X-ray images with spots of color where there may be lung damage or other signs of pneumonia. It has now been applied to more than 6,000 chest X-rays, and its providing some value in diagnosis, said Hsiao, the director of UCSDs augmented imaging and artificial intelligence data analytics laboratory.

His team is one of several around the country that has pushed AI programs into the COVID-19 crisis to perform tasks like deciding which patients face the greatest risk of complications and which can be safely channeled into lower-intensity care.

The machine-learning programs scroll through millions of pieces of data to detect patterns that may be hard for clinicians to discern. Yet few of the algorithms have been rigorously tested against standard procedures. So while they often appear helpful, rolling out the programs in the midst of a pandemic could be confusing to doctors and dangerous for patients, some AI experts warn.

AI is being used for things that are questionable right now, said Dr. Eric Topol, director of the Scripps Research Translational Institute and author of several books on health IT.

Newsletter

Get our free Coronavirus Today newsletter

Sign up for the latest news, best stories and what they mean for you, plus answers to your questions.

You may occasionally receive promotional content from the Los Angeles Times.

Topol singled out a system created by Epic, a major vendor of electronic health records software, that predicts which coronavirus patients may become critically ill. Using the tool before it has been validated is pandemic exceptionalism, he said.

Epic said the companys model had been validated with data from more 16,000 hospitalized COVID-19 patients in 21 healthcare organizations. No research on the tool has been published for independent researchers to assess, but in any case, it was developed to help clinicians make treatment decisions and is not a substitute for their judgment, said James Hickman, a software developer on Epics cognitive computing team.

Others see the COVID-19 crisis as an opportunity to learn about the value of AI tools.

My intuition is its a little bit of the good, bad and ugly, said Eric Perakslis, a data science fellow at Duke University and former chief information officer at the Food and Drug Administration. Research in this setting is important.

Nearly $2 billion poured into companies touting advancements in healthcare AI in 2019. Investments in the first quarter of 2020 totaled $635 million, up from $155 million in the first quarter of 2019, according to digital health technology funder Rock Health.

At least three healthcare AI technology companies have made funding deals specific to the COVID-19 crisis, including Vida Diagnostics, an AI-powered lung-imaging analysis company, according to Rock Health.

Overall, AIs implementation in everyday clinical care is less common than hype over the technology would suggest. Yet the coronavirus has inspired some hospital systems to accelerate promising applications.

UCSD sped up its AI imaging project, rolling it out in only two weeks.

Hsiaos project, with research funding from Amazon Web Services, the University of California and the National Science Foundation, runs every chest X-ray taken at its hospital through an AI algorithm. While no data on the implementation has been published yet, doctors report that the tool influences their clinical decision-making about a third of the time, said Dr. Christopher Longhurst, UCSD Healths chief information officer.

The results to date are very encouraging, and were not seeing any unintended consequences, he said. Anecdotally, were feeling like its helpful, not hurtful.

AI has advanced further in imaging than in other areas of clinical medicine because radiological images have tons of data for algorithms to process, and more data makes the programs more effective, Longhurst said.

But while AI specialists have tried to get AI to do things like predict sepsis and acute respiratory distress researchers at Johns Hopkins University recently won a National Science Foundation grant to use it to predict heart damage in COVID-19 patients it has been easier to plug it into less risky areas such as hospital logistics.

In New York City, two major hospital systems are using AI-enabled algorithms to help them decide when and how patients should move into another phase of care or be sent home.

At Mount Sinai Health System, an artificial intelligence algorithm pinpoints which patients might be ready to be discharged from the hospital within 72 hours, said Robbie Freeman, vice president of clinical innovation at Mount Sinai.

Freeman described the AIs suggestion as a conversation starter, meant to help assist clinicians working on patient cases decide what to do. AI isnt making the decisions.

NYU Langone Health has developed a similar AI model. It predicts whether a COVID-19 patient entering the hospital will suffer adverse events within the next four days, said Dr. Yindalon Aphinyanaphongs, who leads NYU Langones predictive analytics team.

The model will be run in a four- to six-week trial with patients randomized into two groups: one whose doctors will receive the alerts, and another whose doctors will not. The algorithm should help doctors generate a list of things that may predict whether patients are at risk for complications after theyre admitted to the hospital, Aphinyanaphongs said.

Some health systems are leery of rolling out a technology that requires clinical validation in the middle of a pandemic. Others say they didnt need AI to deal with the coronavirus.

Stanford Health Care is not using AI to manage hospitalized patients with COVID-19, said Ron Li, the centers medical informatics director for AI clinical integration. The San Francisco Bay Area hasnt seen the expected surge of patients who would have provided the mass of data needed to make sure AI works on a population, he said.

Outside the hospital, AI-enabled risk factor modeling is being used to help health systems track patients who arent infected with the coronavirus but might be susceptible to complications if they contract COVID-19.

At Scripps Health, clinicians are stratifying patients to assess their risk of getting COVID-19 and experiencing severe symptoms using a risk-scoring model that considers factors like age, chronic conditions and recent hospital visits. When a patient scores 7 or higher, a triage nurse reaches out with information about the coronavirus and may schedule an appointment.

Though emergencies provide unique opportunities to try out advanced tools, its essential for health systems to ensure doctors are comfortable with them, and to use the tools cautiously, with extensive testing and validation, Topol said.

When people are in the heat of battle and overstretched, it would be great to have an algorithm to support them, he said. We just have to make sure the algorithm and the AI tool isnt misleading, because lives are at stake here.

Gold writes for Kaiser Health News, an editorially independent program of the Kaiser Family Foundation. It is not affiliated with Kaiser Permanente.

See the article here:
Coronavirus puts artificial intelligence to the test - Los Angeles Times

May 22, 2020

Two University of Manitoba researchers have received support from Atomwise, the leader in using artificial intelligence (AI) for small molecule drug discovery, to explore broad-spectrum therapies for COVID-19 and other coronaviruses.

Jorg Stetefeld: It is crucial to gain a molecular understanding of how one particularly attractive protein target, nsp12, interacts with another key protein named nsp8. Once learned, this knowledge can be used to develop both new drugs and repurpose existing ones.

Faculty of Science professor Jrg Stetefeld (chemistry), Tier-1 Canada Research Chair in Structural Biology and Biophysics, and associate professor Mark Fry (biological sciences) received support through Atomwises Artificial Intelligence Molecular Screen (AIMS) awards program, which seeks to democratize access to AI for drug discovery and enable researchers to accelerate the translation of their research into novel therapies.

The current pandemic of COVID-19 is caused by a novel virus strain of SARS-CoV-2, says Stetefeld. To develop the most efficient therapeutic strategies to counteract the SARS-CoV-2 infection, it is crucial to gain a molecular understanding of how one particularly attractive protein target, nsp12, interacts with another key protein named nsp8. Once learned, this knowledge can be used to develop both new drugs and repurpose existing ones.

Professro Ben Bailey-Elkin, from the Stetefeld laboratory, will test compounds that Atomwises AI team sends him after they perform an in silico screen of millions of compounds, and carry out the subsequent biochemical and biophysical characterization, significantly reducing the time it would traditionally take to carry out this process. The Atomwise team will use their proprietary AI software to search for promising direct-acting antivirals, which interfere with the function of the viruss targeted proteins.

Professor Frys laboratory will take advantage of Atomwises cutting edge AI to screen a panel of small molecules predicted to interfere with the cellular signaling pathway that is central to the cytokine storm associated with the development of the COVID-19 acute respiratory distress syndrome.

Professor Frys laboratory will take advantage of Atomwises cutting edge AI to screen a panel of small molecules predicted to interfere with the cellular signaling pathway that is central to the cytokine storm.

Cytokines are a group of small proteins secreted by cells for the purpose of cell-to-cell communication, and in healthy individuals, these cytokines regulate key activities such as immunity, cell growth and tissue repair, for example, says Fry. A large number of patients with COVID-19 will develop life threatening pneumonia, accompanied by a so-called cytokine storm where the body experiences excessive or uncontrolled release of a number of these molecules.

Fry adds, The cytokine storm is thought to play a major role in the development of COVID-19, and there is some evidence that drugs which inhibit key cytokines such as interleukin-6 may reduce the severity of the disease. Its important to note that many of these inhibitors are part of a therapeutic class called biological drugs. These can be expensive to make and supply may be limited. My hope is that we can develop a small molecule inhibitor of the cytokine storm that will be easy to synthesize and available to all who need it.

Atomwises patented AI technology has been proven in hundreds of projects to discover drug leads for a wide variety of diseases said Dr. Stacie Calad-Thomson, vice president and head of Artificial Intelligence Molecular Screen (AIMS) Partnerships at Atomwise. Were hopeful that the therapies discovered will not only target this pandemic, but potential future pandemics.

Research at the University of Manitoba is partially supported by funding from the Government of Canada Research Support Fund.

UM Today Staff

Originally posted here:
UM partners with artificial intelligence leader Atomwise to pursue COVID-19 therapies - UM Today

Newswise Sedentary behaviours, poor sleep and questionable food choices are major contributors of chronic disease, including diabetes, anxiety, heart disease and many cancers. But what if we could prevent these through the power of smart technologies?

In a new University of South Australia research project announced today and funded by $1,118,593 from the Medical Research Future Fund (MRFF), researchers will help Australians tackle chronic disease through a range of digital technologies to improve their health.

Using apps, wearables, social media and artificial intelligence, the research will show whether technology can modify and improve peoples behaviours to create meaningful and lasting lifestyle changes that can ward off chronic disease.

Chronic disease is the leading cause of illness, disability and death in Australia with about half of Australians having a least one of eight major conditions including CVD, cancer, arthritis, asthma, back pain, diabetes, pulmonary disease and mental health conditions.

Nearly 40 per cent of chronic disease is preventable through modifiable lifestyle and diet factors.

The research will assess the ability of digital technologies to improve the health and wellbeing across a range of populations, health behaviours and outcomes, with a specific focus on how they can negate poor health outcomes associated high-risk events such as school holidays or Christmas (when people are more likely to indulge and less likely to exercise); where technology could better track the activity among hospital inpatients, outpatients and home-patients (to help recovery from illness and surgery, leading to improved patient outcomes); and how new artificial intelligence-driven virtual health assistants can improve boost health among high-risk groups, such as older adults.

Lead researcher, UniSAs Associate Professor Carol Maher says the research aims to deliver accessible and affordable health solutions for all Australians.

Poor lifestyle patterns a lack of exercise, excess sedentary behaviour, a lack of sleep and poor diets are leading modifiable causes of death and disease in Australia, Assoc Prof Maher says.

Technology has a huge amount to offer in terms of improving lifestyle and health, especially in terms of personalisation and accessibility, but it has to be done thoroughly and it has to be done well.

Research plays an important role in helping understand the products that are most effective, which will see us working with existing commercial technologies and applying and testing them in a new way, as well as developing bespoke software for specific, unmet needs.

The great advantage of technology-delivered programs is that with careful design, once they are developed and evaluated, they can be delivered very affordably and on a massive scale.

If we are to make any change in the prevalence of chronic disease in Australia, we must plan to do it en masse.

The research aims to bridge the gap between academic rigour and commercial offerings so ensure that every Australian has access to the health supports they need.

One of the challenges we face is that many people who could benefit from digital health technologies are intimidated by them for example, older adults who are not that comfortable with technology, or health professionals who are just used to doing things a certain way, Assoc Prof Maher says.

Change can be hard, but when were making leaps in the right direction to improve lifestyle and health of the Australian community, these changes are worth considering.

Read more here:
Harness artificial intelligence and take control your health - Newswise