Surprise Billing Spotlight | Brazil Scandal Goes Global | Responsible ML

“you have the power to stop insurance companies’ rate setting scheme.”

A social media ad from pro-physician PAC, Doctor Patient Unity, addressing dozens of U.S. House and Senate members across the country and pushing for a solution that makes insurers “pay their fair share.”

Imaging Wire Sponsors

  • Carestream – Focused on delivering innovation that is life changing – for patients, customers, employees, communities and other stakeholders
  • Focused Ultrasound Foundation – Accelerating the development and adoption of focused ultrasound
  • Medmo – Helping underinsured Americans save on medical scans by connecting them to imaging providers with unfilled schedule time
  • Nuance – AI and cloud-powered technology solutions to help radiologists stay focused, move quickly, and work smarter
  • Pocus Systems – A new Point of Care Ultrasound startup, combining a team of POCUS veterans with next-generation genuine AI technology to disrupt the industry
  • Qure.ai – Making healthcare more accessible by applying deep learning to radiology imaging

The Imaging Wire

Surprise Billing in the Spotlight

The battle over surprise medical billing hit a fever pitch in recent weeks, as industry players (providers, insurers, the media) ramped up their efforts ahead of Congress’ decision whether surprise billing (and physician reimbursement) will be settled through arbitration or by using median benchmark rates.

  • The Influence Machine – Groups on the provider side of the industry like Physicians for Fair Coverage and Doctor Patient Unity are spending big on TV and online ads that position benchmarking as an insurer-created scheme that will hurt physicians and patients, while directly targeting local Congress people in each region.
  • RP Speaks UpRadiology Partners weighed in on surprise medical billing, acknowledging that balance bills are bad and agreeing that patients shouldn’t pay more for unanticipated out-of-network care, while stating a similar case against the benchmark approach. RP likened the benchmark approach to “federal price-fixing,” arguing that creating a standard for out-of-network payments would set a “new ceiling for in-network payments” and cause insurers to cancel their contracts if out-of-network rates are lower. Most importantly, RP urged radiologists to educate themselves on this issue and tell their legislators where they stand.
  • The Press Takes Notice – The media has officially picked up on the surprise billing story, covering the surge in surprise billing political ads, the pros and cons of each option, and what to expect from Congress. The New York Time’s surprise billing guru, Sarah Kliff (formerly VOX, always excellent), went a step further publishing a JAMA editorial on her findings from VOX’s database of ER bills.

Brazil Scandal Goes Global

Reuters’ investigation into the Brazilian medical device corruption scandal took a global twist last week with new allegations that Philips’ local and global leaders were warned of suspicious government deals (bid rigging, kickbacks, price fixing), but didn’t address the issue. Here’s the main takeaways:

  • The Global Twist – Now in its second year (here’s some highlights from 2018 and 2019), the Brazil scandal has only brought down manufacturers’ local employees/executives and middlemen, while keeping a safe distance from their global teams. However, the whistleblower (a former manager at a Philips Brazil subsidiary) revealed that reports of “malfeasance reached the highest levels of the Dutch conglomerate as early as 2010.”
  • Whistleblower – The ex-employee emailed a Philips hotline to report his suspicions about three deals with “an obscure Brazilian middleman” who won large contracts with Brazil’s Ministry of Health (defibrillators, vital-sign monitors). He met with Philips’ top compliance officer and also alerted at least three other senior executives, including former Philips Healthcare CEO Steve Rusckowski.
  • Philips’ Response – Philips confirmed that it launched an internal investigation into the deal but “did not identify direct evidence of wrongdoing.” Although Philips didn’t pull out of the deal, the company did strengthen its internal control processes in Brazil.
  • Philips isn’t Alone – A number of major imaging and medtech brands (Siemens, GE, J&J, Stryker) have also been named in the Brazil scandal. However, the Philips whistleblower is the first to tie allegations to global executives, and he’s cooperating with prosecutors.

Responsible ML Roadmap

A new perspective in Nature Magazine outlined a seven-step roadmap to responsible healthcare machine learning. Here’s the roadmap:

  • Choose the Right Problem – Rather than target healthcare problems based on the fact that an annotated ML training database already exists for it, develop solutions that tackle problems that need to be solved and position them for use by people who can solve them. The article suggested that ML researchers engage stakeholders (clinicians, patients, admin leaders, insurers) early in any project to identify the right problem to focus on.
  • Develop a Useful Solution – Once a problem is ID’d and data access is established, scrutinize the data to make sure it fits the targeted problem (both from an input and output perspective) and to ensure that the data is harmonized and representative of the environment it will be used in.
  • Consider Ethical Implications – Inspect for potential biases in the data and explore how to mitigate their effect. Keep in mind that there are ethical challenges throughout the ML process, and consider working with ethicists to ensure privacy, safety, and fair treatment of all patients.
  • Rigorously Evaluate the Model – There are numerous ways models can go wrong, making it crucial to evaluate their accuracy and value.
  • Thoughtfully Report Results – It’s important to be mindful when interpreting and reporting results, including clearly describing the data, participants, outcomes and predictors, and sometimes even presenting the model.
  • Deploy Responsibly – Before bringing a model into patient care, the system should be tested in ‘silent’ mode (predictions are made, but not acted upon), then moving on to clinical studies, then carefully integrating it into clinical workflow, and then continuously monitoring and improving the solution.
  • Make it to Market – Navigate the regulatory process based on the requirements/preferences of each regulatory body, noting that interpretable ML may be preferred, and working with regulatory experts is suggested.

Although the article is clearly positioned as a guideline for “doing no harm” with AI, this is a valuable roadmap for anyone in healthcare AI. Things like solving the right problems and making sure a solution works are important no matter how ‘woke’ you are.

The Wire

  • An FDA Radiological Health team analyzed 1,568 adverse MRI event reports from 2008 to 2017, finding that most events were preventable. Thermal events were the common cause of MRI-related injuries (59%), followed by mechanical events (11%), projectile events (9%), and acoustic events (6%).
  • IHS Markit revealed that point-of-care ultrasound needle guidance applications drove 15% growth in the POCUS for anesthesia segment and 10% growth in the POCUS for MSK segment in 2018. The firm expects continued growth through 2023 with U.S. demand helped by ultrasound needle guidance’s role in combating the opioid crisis, European demand driven by POCUS’ cost benefits and effectiveness, and Asia demand propelled by joint replacement surgery applications.
  • Feeling uncertain about how you’re going to adopt an effective appropriate imaging system by 2021 (or PAMA’s 2020 soft launch)? A new piece from MedPage Today reveals that you’re not alone, detailing providers’ concerns about the work required to launch clinical decision support software and doubts whether CDSM software will actually result in more appropriate imaging orders (this study wasn’t very bullish). As a result, imaging societies are pushing CMS to further postpone PAMA until they can find a better way to “apply AUC and practice value-based imaging.”
  • Researchers from Washington University (the St. Louis one) and the University of Birmingham (the UK one) found that advancements in near infrared spectroscopy (fNIRS) could improve brain imaging. The researchers used a high-density diffuse optical tomography (HD-DOT) configuration (allows overlapping measurements at multiple source-detector distances) that improved image resolution (by up to 21%), allowed greater depth sensitivity, and reduced frequency-domain (up to 59%).
  • GE is having an eventful few weeks, starting on the 16th when the guy who discovered Bernie Madoff’s Ponzi scheme alleged that GE is committing “bigger than Enron fraud.” Wall Street reacted with a massive 11% drop in GE’s stock value, leading to GE executives making major stock purchases and issuing public rebuttals, prompting a decent 9.7% stock rebound. Accounting expert Harry Markopolos criticized how GE accounts for its Baker Hughes oil-and-gas business and its long-term-care insurance holdings, accusing the company of obscuring its financial problems and filing inaccurate or fraudulent information with regulators.
  • Research published in JAMA Network Open (n= 1,246 patients) reveals that 67.1% of patients would share their medical records and biospecimens for research purposes with their home institution and 25% would share their info with all researchers (including other institutions), while only 3.7% were unwilling to share any info. However, most of these patients would prefer to know how their information would be used and some would prefer to withhold specific info or only share their info with certain institutions (47.4% weren’t open to sharing with for-profit institutions).
  • Here’s another ‘AI beats radiologists’ story for everyone. An algorithm from Chinese AI startup, JF Healthcare, was the first to beat a team of Stanford radiologists in the university’s CheXpert chest X-ray interpretation competition. The JF Healthcare algorithm achieved a 0.926 average AUC score, beating three radiologists, and giving it the top spot on the Stanford CheXpert leaderboard.
  • A Pennsylvania couple was awarded an $8.5 million malpractice judgement against St. Luke’s Health Network after a pair of doctors did not effectively inform the husband of a potential tumor in his bladder. The gentleman visited a St. Luke’s emergency room in 2015 with pain in his right side and abdomen and urinary problems, and although a radiology report identified kidney stones and a potential tumor, he wasn’t informed of the possible tumor or told to seek further evaluation. A 2017 bladder ultrasound revealed he had multiple tumors and bladder cancer.
  • Feedback plc raised £2 million ($2.455m) to support the upcoming launch of its Bleepa app, which expands upon the company’s Cadran PACS system, allowing “medical grade” image messaging and viewing on mobile devices. The funding round follows a £1.375 million ($1.688M) round in late 2018 that it used to expand its Cadran PACS customer base/customer support, fund the FDA clearance process for its TexRAD texture analysis software, and develop its U.S. partnership with Imaging Endpoints.
  • GE Healthcare India announced its new Edison[X] program, supporting select Indian healthcare startups with a focus on companies in the healthcare IOT, AI/ML, and big data/advanced analytics spaces. GE will provide the startups with $10k zero-equity grants to develop their solutions and provide them with mentorship, skill development, commercial and regulatory guidance, and data provision.
  • An editorial from Radiology: Artificial Intelligence’s Charles E. Kahn, Jr, MD, MS stated the case for explainable AI, noting that beyond AI activation maps (heat maps that show which parts of an image are being used by an AI model), there are few ways for humans to understand an AI tool’s reasoning so they can interpret its output. Dr. Kahn added that AI systems can learn patterns but can’t construct “mental models” to connect their learning with concepts that humans can understand and therefore trust.

The Resource Wire

– This is sponsored content.

  • This Carestream case study compares images of foot trauma captured using the OnSight 3D Extremity System to images captured on 2D X-rays.
  • Nuance’s latest blog highlights how ImageBiopsy Lab’s KOALA (Knee Osteoarthritis Labeling Assistant) algorithm helps physicians spot signs of knee osteoarthritis.
  • Yale University research reveals that the average patient drives past SIX lower-cost providers on the way to an imaging procedure, due in large part to patients’ and physicians’ limited cost consciousness. Medmo helps address this issue by letting patients enter what they can afford for their scan, then booking them at a nearby imaging center willing to accept that price.

You might also like

You might also like..

Select All

You're signed up!

It's great to have you as a reader. Check your inbox for a welcome email.

-- The Imaging Wire team

You're all set!