Perspectives and Best Practices for Artificial Intelligence and Continuously Learning Systems in Healthcare

Healthcare is often a late adopter when it comes to new techniques and technologies; this works to our advantage in the development of this paper as we relied on lessons learned from CLS in other industries to help guide the content of this paper. Appendix V includes a number of example use cases of AI in Healthcare and other industries.

This paper focuses on identifying unique attributes, constraints and potential best practices towards what might represent “good” development for Continuously Learning Systems (CLS) AI systems with applications ranging from pharmaceutical applications for new drug development and research to AI enabled smart medical devices. It should be noted that although the emphasis of this paper is on CLS, some of these issues are common to all AI products in healthcare.

The full paper can be downloaded below.  

Electronic Health Records as a GPS for Healthcare

As patients are bombarded with more choice and information than ever, the burdened health system seems to lack the appropriate support to manage increasing demands for personalized and convenient care.

Today’s infographic comes to us from Publicis Health, and it demonstrates how electronic health records are an important piece in the puzzle to improve experiences for patients and providers alike.

The full infographic and article can be viewed at this link.  

Deprescribing interventions and their impact on medication adherence in community-dwelling older adults with polypharmacy: a systematic review

This systematic review found that deprescribing as an intervention did not routinely improve medication adherence in this patient population. The theory that a reduced medication burden would improve adherence could not be substantiated in the literature. This is because the interventions described in the studies did not convincingly reduce medication burden. There is a range of bio-psycho-social factors reported that associate with improved adherence, but medicines review processes vary and rarely report the population demographic. Adherence is mostly reported as a secondary outcome and there is no standard report of successful adherence to medication. The authors recommend further study into the efficacy and outcomes of medicines management interventions. A consensus on priority outcome measurements for prescribed medications is indicated.

The full article can be downloaded below.  

Association of Pharmaceutical Industry Marketing of Opioid Products With Mortality From Opioid-Related Overdoses

Prescription opioids are involved in 40% of all deaths from opioid overdose in the United States and are commonly the first opioids encountered by individuals with opioid use disorder. It is unclear whether the pharmaceutical industry marketing of opioids to physicians is associated with mortality from overdoses.

The objective was to identify the association between direct-to-physician marketing of opioid products by pharmaceutical companies and mortality from prescription opioid overdoses across US counties.

This population-based, county-level analysis of industry marketing information used data from the Centers for Medicare & Medicaid Services Open Payments database linked with data from the Centers for Disease Control and Prevention on opioid prescribing and mortality from overdoses. All US counties were included, with data on overdoses from August 1, 2014, to December 31, 2016, linked to marketing data from August 1, 2013, to December 31, 2015, using a 1-year lag. Statistical analyses were conducted between February 1 and June 1, 2018.

The main outcomes and measures were county-level mortality from prescription opioid overdoses, total cost of marketing of opioid products to physicians, number of marketing interactions, opioid prescribing rates, and sociodemographic factors.

Between August 1, 2013, and December 31, 2015, there were 434 754 payments totaling $39.7 million in nonresearch-based opioid marketing distributed to 67 507 physicians across 2208 US counties. After adjustment for county-level sociodemographic factors, mortality from opioid overdoses increased with each 1-SD increase in marketing value in dollars per capita (adjusted relative risk, 1.09; 95% CI, 1.05-1.12), number of payments to physicians per capita (adjusted relative risk, 1.18; 95% CI, 1.14-1.21, and number of physicians receiving marketing per capita (adjusted relative risk, 1.12; 95% CI, 1.08-1.16). Opioid prescribing rates also increased with marketing and partially mediated the association between marketing and mortality.

In this study, across US counties, marketing of opioid products to physicians was associated with increased opioid prescribing and, subsequently, with elevated mortality from overdoses. Amid a national opioid overdose crisis, reexamining the influence of the pharmaceutical industry may be warranted.

The full article can be downloaded below.  

Error types with use of medication-related technology: A mixed methods research study

Health information technology has been integrated throughout the medication use process to enhance safety, quality, and care efficiency. However, technologies have the potential to eliminate or reduce, but also create some new types of errors.

The objective was to ssess specific error types before and after the incorporation of two different health information technologies (HITs), e-prescribing and automated dispensing cabinets (ADCs), into pharmacists’ daily work.

A mixed methods design guided use of a pre-existing database of pharmacist survey responses describing patient safety HIT-related issues in the form of errors prevented and errors observed. In vivo descriptive text responses were converted into error types. Descriptive analysis was performed to characterize the error types associated with each HIT.

Four error types were eliminated with the use of e-prescribing, three new error types emerged, and three error types persisted. With ADC use, four error types were eliminated, three new error types emerged, and three error types persisted.

Each technology has its own error types, and some persist regardless of HIT use. There is a need to determine optimal risk reduction approaches for each unique HIT introduced, and design safety practice improvement for error types unaffected by the introduction of HIT use.

The full article can be downloaded below.  

Assessing the Unintended Consequences of Health Policy on Rural Populations and Places 

Because of the complexity of the U.S. health care system, thoughtfully designed health policies carry a risk of having unintended consequences, particularly for health systems in rural places that have place-based fundamentals that deviate substantially from urban and suburban areas. Policies developed without consideration of rural contexts are likely to create unanticipated and negative consequences for rural residents, providers, and communities.

When health policies are being developed, a number of themes that emerge are useful to keep in mind. Specifically, how will this policy impact the ability of a rural health system to offer essential, affordable, and high-quality services to rural populations? How might this policy result in disparate outcomes and widen health inequities, such as threatening access, slowing quality improvement, or creating financial barriers to obtaining health insurance or buying health care services?

The rural-proofing framework presented in this paper is a policy analysis tool for thinking about what the unintended consequences of a policy may be on rural populations and places vis-à-vis the objectives of a high-performance rural health system. Policy analysis must be applied to all sources of authoritative actions given that policies are produced not just in the legislative context, but also through judicial, administrative, and rulemaking actions.

The full report can be downloaded below.  

Rapid detection of internalizing diagnosis in young children enabled by wearable sensors and machine learning

There is a critical need for fast, inexpensive, objective, and accurate screening tools for childhood psychopathology. Perhaps most compelling is in the case of internalizing disorders, like anxiety and depression, where unobservable symptoms cause children to go unassessed–suffering in silence because they never exhibiting the disruptive behaviors that would lead to a referral for diagnostic assessment. If left untreated these disorders are associated with long-term negative outcomes including substance abuse and increased risk for suicide. This paper presents a new approach for identifying children with internalizing disorders using an instrumented 90-second mood induction task. Participant motion during the task is monitored using a commercially available wearable sensor. We show that machine learning can be used to differentiate children with an internalizing diagnosis from controls with 81% accuracy (67% sensitivity, 88% specificity). We provide a detailed description of the modeling methodology used to arrive at these results and explore further the predictive ability of each temporal phase of the mood induction task. Kinematical measures most discriminative of internalizing diagnosis are analyzed in detail, showing affected children exhibit significantly more avoidance of ambiguous threat. Performance of the proposed approach is compared to clinical thresholds on parent-reported child symptoms which differentiate children with an internalizing diagnosis from controls with slightly lower accuracy (.68-.75 vs. .81), slightly higher specificity (.88–1.00 vs. .88), and lower sensitivity (.00-.42 vs. .67) than the proposed, instrumented method. These results point toward the future use of this approach for screening children for internalizing disorders so that interventions can be deployed when they have the highest chance for long-term success.

The full article can be downloaded below.  

TEMPE, Ariz. (Dec. 10, 2018) – Sonora Quest Laboratories, Arizona’s leading provider of diagnostic testing and information services, and Beacon Biomedical Inc., a Phoenix-based diagnostic laboratory focused on commercializing early detection cancer tests, have jointly announced the availability of Beacon’s new colorectal cancer screening blood test called BeScreened™-CRC. Under an exclusive partnership the test is now available directly to consumers without a doctor’s order through Sonora Quest’s My Lab ReQuest™, their self-ordered lab testing product which is available on-line and at all of Sonora Quest’s 70-plus patient service centers across Arizona. In the near future, Sonora Quest and Beacon Biomedical intend on offering BeScreened-CRC throughout their entire network, including primary healthcare providers.

Recognizing that many people are unable or unwilling to use existing screening tests, Beacon Biomedical developed BeScreened-CRC as a convenient alternative to enable early detection, improve healthcare outcomes and reduce costs. BeScreened-CRC is 94 percent accurate at detecting the likely presence or absence of colorectal cancer, even in its earliest stages. As a blood test, it has the potential to remove many of the barriers to screening due to the test’s simplicity and ease of adoption, and it can be collected at the same time as other blood tests with no special preparation needed.

Download the press release to read more!

TEMPE, Ariz. (Nov. 9, 2018) – Sonora Quest Laboratories, Arizona’s leading provider of diagnostic testing and information services, and BaseHealth, a predictive health analytics company that empowers health systems and Accountable Care Organizations (ACO’s) to proactively manage rising risk within populations, are collaborating to offer a pioneering analytics platform that identifies unknown risks in patient populations for improved outcomes and presents opportunities for cost savings for healthcare systems across the country.

Download the press release to read more.

October 2018 whitepaper from Sonora Quest Laboratories and Commonwealth Primary Care ACO.

ACO success requires current and relevant data access. Data in healthcare is often siloed, challenging to aggregate, and difficult to act on. In order to be effective in this new healthcare landscape, Sonora Quest Laboratories (SQL) and Commonwealth Primary Care ACO (Commonwealth) worked together to analyze information from their respective patient databases in order to assist Commonwealth in serving their patients.

This paper demonstrates how the strategic partnership between Commonwealth and SQL resulted in actionable patient data, improved communications with providers, and also led to the development of data-focused care models utilizing SQL’s Actionable Insights Management™ (AIM™) analytics platform.