Machine Learning and AI Transform Healthcare Data Management so Healthcare Stakeholders Can Achieve Lower Costs and Better Patient Outcomes

Healthcare is a rapidly changing industry. Big data, artificial intelligence (AI), and machine learning – these terms have become mainstream in many industries and healthcare is no exception. AI applications are increasingly becoming a part of healthcare to the point where there have been rumblings about AI possibly taking over some healthcare jobs. But is this realistic? Should healthcare stakeholders all just embrace AI in healthcare?

What is Artificial Intelligence (AI)?

Artificial intelligence is a collective term used in reference to multiple technologies that aim to mimic human cognitive functions, making machines able to sense, understand, act, and learn. Some AI technologies that are important to healthcare stakeholders and the healthcare industry are:

• Machine learning – neural networks and deep learning: statistical techniques used to fit models to data and to ‘learn’ by training models with data.
• Natural language processing (NLP): making sense of human language with applications such as speech recognition, text analysis, and translation.
• Rule-based expert systems: based on “if-then” rules.
• Physical robots: actual robots such as surgical robots.
• Robotic process automation: computer programs used to perform structured digital tasks for administrative purposes.

AI in Healthcare – Past, Present, Future

Healthcare is a data rich industry which provides fertile ground for applications of AI in healthcare data management as well as other aspects of healthcare. In the past, AI technologies in healthcare were mainly algorithms or tools that complement a human. The rule-based expert systems of “if-then” rules were dominant in the 1980’s and for some time beyond. However, inefficiencies in such systems that result when the number of rules become too large, are causing them to be replaced by machine learning algorithms.

Today, AI in healthcare can truly augment human activity and is taking over tasks ranging from medical imaging to risk analysis to diagnosing health conditions. AI is being used to discover links between genetic codes, to power surgical robots, and to maximize hospital efficiency. As a result of this, growth of AI in healthcare is exploding and is estimated to reach $6.6 billion by 2021. An analysis of the market found that when combined, key clinical health AI applications can potentially create $150 billion in annual savings for the United States healthcare economy by 2026.

In the future, AI will play a critical role in precision medicine, which is the direction in which healthcare delivery is headed. AI is expected to improve areas such as diagnosis and providing treatment recommendations, imaging analysis, healthcare data management, patient communication, and capturing of clinical notes through speech and text recognition.

According to a PWC report, Why AI and Robotics will Define New Health, AI in past decades have focused on innovations in medical products (equipment, hardware, and software) that deliver historic and evidence-based care. The present decade has seen a rise in medical platforms focused on real-time, outcome-based care in the form of wearables, big data, healthcare data management, and health analytics. Healthcare stakeholders should be aware that it is believed that the next decade will see an increase in medical solutions involving robotics and augmented reality, which will deliver intelligent solutions for both evidence and outcome-based health and focus on collaborative, preventative care.

Uses of AI in Healthcare

“AI is being used to discover links between genetic codes, to power surgical robots, and to maximize hospital efficiency.”

One description of AI is that it “simplifies the lives of patients, doctors and hospital administrators by performing tasks that are typically done by humans, but in less time and at a fraction of the cost.” AI technologies have found a multitude of uses in healthcare such as:

• Efficiently and accurately diagnosing conditions and reducing errors
• Treatment recommendations
• Development of medicines
• Streamlining the patient experience
• Data mining and healthcare data management
• Robot-assisted surgery
• Administrative applications

With all these uses and more, AI in healthcare is here to stay and healthcare stakeholders who want to remain relevant and thrive in this space, need to make the necessary moves now to incorporate AI into their operations and optimize healthcare data management.

Impact of AI on Healthcare Data Management and Healthcare Stakeholders

The amount of data available in healthcare is staggering. Data sources include electronic medical records (EMRs), insurance claims, clinical trials, drug research and development, and patient generated health data. All these points of data have the potential to change the healthcare landscape if properly managed and leveraged accordingly. However, without proper healthcare data management, valuable data can become lost among the large volumes of data points available; this is where AI can help healthcare stakeholders.

“Healthcare could save up to $100 billion a year by utilizing big-data crunching algorithms backed by AI to inform decision-making and realize efficiencies in clinical trials and research”

It is estimated that the industry could save up to $100 billion a year by utilizing big-data crunching algorithms backed by AI to inform decision-making and realize efficiencies in clinical trials and research. There are companies that are leveraging the power of AI to assist healthcare stakeholders with healthcare data management and improving healthcare delivery. Tempus is one company that is using AI tools to collect and analyze the world’s largest library of clinical and molecular data to drive precision medicine. Its AI-driven data are being used in cancer research and treatment.

KenSci is another company that is using the power of AI and big data to improve healthcare. Their risk reduction platform aggregates data from existing sources such as EMRs, claims, and financial data to help uncover clinical, operational, and financial risks. It can predict who might get sick and the drivers behind healthcare costs; it can also provide solutions to these problems.

H2O.ai uses AI to analyze data throughout a healthcare system to mine, automate, and predict processes. It has been used to foresee ICU transfers, improve clinical workflows, and even pinpoint a patient’s risk of hospital-acquired infections.

AI-enabled applications are also impacting healthcare data management practices. Computer-assisted coding (CAC) is on the rise, utilizing NLP to read and interpret clinical documentation in patient health records and suggest applicable diagnosis and procedure codes. For CAC to be fully adopted and optimal efficiencies realized, the medical coding workflow will have to be re-engineered. As machine learning becomes more integral to reading images for diagnosis, the requirement for a physician to interpret an image may become less necessary. As such, medical coding and reporting guidelines and standards will need to be adjusted to account for AI applications.

Benefits of AI in Healthcare

AI has proven to be a boon to the healthcare industry. Some of the benefits from incorporating AI technologies in healthcare include:

• Faster, earlier, and more accurate diagnoses
• More efficient data mining and healthcare data management
• Lowered costs
• Improved patient outcomes
• Enhanced patient engagement
• Healthier behaviors and proactive lifestyle management with wearable technology
• Better understanding of the patient’s condition and improved management resulting from increased insight of the healthcare team into the day to day lives patients
• Improved clinical decision-making
• Reduced administrative burdens and improved efficiency in managing administrative tasks
• More efficient drug research and discovery process potentially cutting both the time to market for new drugs and their costs significantly

AI technologies are already proving to be a game-changer in healthcare and there is still a huge potential for them to do even more. Many companies including big names such as IBM and Google are investing heavily in AI for healthcare. We are on the cusp of many discoveries and breakthroughs as the potential of AI is fully realized in the healthcare industry.

Challenges/Pitfalls of AI in Healthcare

While the advancement of AI technologies in healthcare is exciting and offers numerous benefits, like everything else, it is not without challenges. Some healthcare stakeholders prefer to take a cautious approach, with some wondering how far is too far. Below are some of the challenges/pitfalls facing healthcare stakeholders and the evolution of AI in healthcare:

• Potential biases in results based on the data used to create algorithms and ‘train’ the AI technology.
• Transferability of algorithms ‘trained’ in one setting or population to another.
• Data ownership, confidentiality, and consent – who owns the data? Who is responsible for healthcare data management? How will patients’ data be kept confidential? How do healthcare stakeholders handle the need for consent in research?
• Ethical issues and professional responsibility – who is responsible when a patient is misdiagnosed?
• Legal risks and regulatory issues – at present, regulations are falling behind the explosion in AI development and use potentially creating a legal and regulatory nightmare for healthcare stakeholders.

Healthcare is experiencing an upsurge in the development and application of AI technologies. These technologies are very beneficial in that they can perform tasks that are usually done by humans in less time and at much lower costs, simplifying the lives of healthcare stakeholders including patients, doctors, and hospital administrators. Healthcare data management is one area that has benefited significantly from AI and has resulted in better patient outcomes and cost savings for healthcare stakeholders. However, there are some serious challenges to the use of AI in healthcare that need to be overcome. While AI might not take over healthcare jobs, these challenges need to be managed before we can all completely embrace the full potential of AI in healthcare.

To find out more about how wearables and patient generated health data are part of the bigger AI picture in healthcare, contact the healthcare technology experts at Acuma Health.

Adverse Drug Reactions vs. Adverse Drug Events – Detect and Avoid for Patient Safety

We are experiencing a health craze. Everybody is becoming health conscious and there is an abundance of herbs, supplements, lotions, etc. that people are using in their quest to rid themselves of, and prevent, ailments.

In addition, there are numerous prescription medications (almost 6,800) and countless over-the-counter drugs available in the US market. The use of so many substances in health care opens the door to possible interactions between substances, resulting in medication adverse effects. In the clinical space, terms such as adverse drug events (ADEs) and adverse drug reactions (ADRs) are used to describe some of the possible medication adverse effects that can result from drug use. But does the average person understand the meaning of these terms?

An adverse drug event (ADE) is defined as “an injury resulting from medical intervention related to a drug.” An ADE results from harms caused directly by the drug itself and include medication errors, ADRs, overdoses, and allergic reactions. An ADR, on the other hand, is a harmful and unintended response to a drug at normal doses and during normal use.

The Office of Disease Prevention and Health Promotion, part of the US Department of Health and Human Services, has named ADE prevention as an important patient safety priority in its National Action Plan for Adverse Drug Event Reporting. It noted that ADEs accounted for an estimated one-third of hospital adverse events and approximately 280,000 hospital admissions annually. The ADE Action Plan was established to coordinate multiple stakeholders and align Federal efforts in identifying common, preventable, and measurable ADEs that may result in significant patient harm. The goal is to jointly work towards reducing patient harm from these specific identified ADEs nationally.

Three types of ADEs were selected as the high-priority targets of the ADE Action Plan as they were identified as being common, clinically significant, preventable, and measurable. These are:

• Anticoagulants: primary ADE of concern – bleeding
• Diabetes agents: primary ADE of concern – hypoglycemia
• Opioids: primary ADE of concern – accidental overdoses/over-sedation/respiratory depression

The World Health Organization lists has reported that as many as 4 in 10 patients globally are harmed in primary and outpatient health care. Up to 80% of harm is preventable. The most detrimental errors are related to diagnosis, prescription and the use of medicines. As such, healthcare organizations are urged to utilize the ADE Action Plan and implement strategies to prevent medication adverse effects, especially from the three priority types of ADEs identified. Healthcare data management and healthcare technology can play a major role in this regard, and those organizations that proactively use healthcare technology to detect ADEs will be a step ahead in preventing them. Some ways in which healthcare data management and healthcare technology can be used in ADE detection and prevention include:

• Electronic exchange of health information, such as laboratory results and care (e.g., discharge) summaries. This can help to improve communication among the care team as a patient passes from one team to the next.
• Interoperability between laboratory and pharmacy systems to help prevent medication errors and medication adverse effects.
• Utilize electronic health records (EHRs) and patient engagement tools to provide patient-specific data that can inform appropriate clinical decisions by providers. EHRs and patient engagement tools can also provide clinical reminders and templates to prompt and facilitate recommended clinical practices. This could result in improvements in assessment, documentation, and collaborative treatment planning for patient risk factors and aberrant behaviors.
• Leverage EHR Meaningful Use requirements by incorporating Quality Measures specific to the three types of ADEs identified in the ADE Action Plan.

ADEs can also be prevented by improving patient compliance with their medication regimen. Incorporating remote monitoring solutions such as automated medication dispensers with reminders can improve patient compliance and also prevent potential overdoses especially in seniors with memory problems.

Patient safety is a priority for all healthcare organizations and medication adverse effects are a serious threat to this. By proactively incorporating healthcare technology to detect and avoid ADEs, organizations can increasingly prevent them and improve patient safety. Acuma Health can show you how to utilize technology for proactive prevention of medication adverse effects and improved patient safety.

Appropriate PGHD Frameworks Need to Be Put in Place to Ensure Healthcare Stakeholders Comply with Regulatory Requirement

The healthcare landscape is changing. Patients are becoming much more interested and invested in their health and care decisions, and their voices and experiences are more important than ever before. Regulators are requiring and/or incentivising the inclusion of the patient’s voice in care delivery. Healthcare providers are increasingly being required to put the patient at the centre of the care they provide, and all healthcare stakeholders must find ways of incorporating the patient’s perspective into their care setting. Healthcare is moving towards what it should have always been about, the patients.

Healthcare Technology and the Rise in Patient Generated Health Data

Facilitated by an explosion in the use of healthcare technology, patients nowadays can collect and keep track of a variety of health indicators such as fitness levels, dietary factors, symptoms, and treatment history, which they can share with their healthcare providers. This patient generated health data has the potential to transform health systems, improving the quality of care and patient outcomes. Patient generated health data (PGHD) as defined by the Office of the National Coordinator for Health Information Technology (ONC) is “health-related data created, recorded, or gathered by or from patients (or family members or other caregivers) to help address a health concern.”

“There are over 318,000 mobile applications available to consumers and more than 340 consumer wearable devices on the market worldwide.”

–IQVIA Institute for Human Data Science Report

While PGHD has long been a part of healthcare in ways such as patients keeping track of their weight on paper or telling their providers about their symptoms, the significant increase being seen in the collection of PGHD is due largely to innovative digital health technologies such as wearables, which have become more popular and affordable. Consumer level devices such as the Fitbit and Smart Monitor’s Inspyre^TM; mobile health (mHealth) applications; and registered medical devices such as blood glucose monitors and interactive weight scales, are all giving the patient the power to easily collect and share their PGHD.

Benefits of Collecting and Sharing Patient Generated Health Data

When combined with existing clinical data, PGHD helps to provide a more fulsome picture of the health of the patient in the everyday environment outside of the care setting, which can be used to better inform care decisions. The collection and sharing of PGHD has benefits for the patient, the provider, and all healthcare stakeholders. These benefits include:

• Better disease management, especially for chronic conditions
• Improved patient outcomes through more timely diagnosis, proactive monitoring of changes in routine, and identification of conditions before they get worse
• Reduced costs due to fewer hospital admissions and readmissions as a result of earlier treatment interventions
• Decreased penalties from lower readmission rates
• Increased patient engagement
• Compliance with federal regulations requiring the incorporation of PGHD in electronic health records (EHRs)
• Improvements in the healthcare system through big data analytics

Leveraging PGHD can change the digital health game, improving analytics and reducing costs.

Challenges to the Collection and Use of Patient Generated Health Data

Despite the many significant benefits to be derived from collecting and using PGHD, it is not without challenges. The ONC in its whitepaper Conceptualizing a Data Infrastructure for theCapture, Use, and Sharing of Patient-GeneratedHealth Data in Care Delivery and Researchthrough 2024, shares some challenges that may be faced by patients, providers, and healthcare systems:

• Patients may not understand the advantages to be had in capturing and sharing their PGHD with providers and researchers
• Patients have different levels of health and technology literacy
• Data privacy and security concerns
• Lack of technical infrastructure, workforce capacity, and training in the healthcare system to adequately manage the intake and analysis of the large volumes of PGHD
• Confirming the validity and accuracy of the PGHD generated by varying devices

Notwithstanding these challenges, the potential and opportunities presented by the proper collection, handling, and use of PGHD in improving care and reducing costs, necessitate the development and implementation of proper frameworks to ensure that PGHD are utilized properly and healthcare stakeholders comply with requisite regulations while incorporating PGHD in their operations.

Proper Frameworks are Critical for Collecting, Storing, and Utilizing Patient Generated Health Data

PGHD are everywhere in the healthcare space. Numerous devices and mHealth applications abound that allow for collecting, storing, and sharing PGHD. Almost all patients are collecting PGHD in some form resulting in large volumes of data. Proper healthcare data management now becomes critical as without this, things get chaotic and the data can become useless. Additionally, data privacy and security issues can lead to non-compliance with regulations and fines. Some pertinent questions to be asked include when seeking to capture and utilize PGHD include:

• How do we properly collect all this data?
• How do we ensure the accuracy of the data being collected?
• How can we separate data that are useful from those that are not?
• How can we turn the raw data into useful information that can guide care decisions?
• Where will all this data be stored? Do we have the capability to store large amounts of data?
• What privacy and security measures do we need to put in place?
• Can PGHD be incorporated into our EHRs?
• How do we ensure that we properly handle and utilize PGHD such that we remain in compliance with relevant regulations?

When developing frameworks for the capture, sharing, and use of PGHD, it is critical to ascertain what federal, state, or organizational laws and regulations are relevant to the process to avoid non-compliance

All these questions can be answered with a proper framework in place. Stemming from two pilot demonstrations to test some concepts on the use of PGHD in real-life situations, the ONC has prepared a practical guide offering best practices and questions to consider when developing a policy framework for capturing, using, and sharing PGHD in clinical and research settings. The guide suggests that healthcare stakeholders should consider the following four areas:

1. Strategic planning: this includes determining the priorities and objectives of the organization, assessing the business case, and securing executive sponsorship and enlisting support at all levels
2. Defining requirements: identifying patient-facing technologies is covered in this section
3. Implementation: this incorporates training staff, recruiting and enrolling patients, and reviewing and acting on the PGHD collected
4. Monitoring and adapting: understanding and adhering to relevant privacy and security laws and regulations is one area should be looked at in this section

Following these steps will help to ensure that a proper structure is in place for leveraging PGHD while remaining compliant with regulations.

Regulations and Requirements on PGHD for Healthcare Stakeholders

One of the main concerns with the capture, sharing, anduse of PGHD, and which has the potential to cause violations to regulations, is privacy and security of the data. The Health Insurance Portability and Accountability Act (HIPAA) Privacy Rule, while not necessarily specific to PGHD, aims to “assure that individuals’ health information is properly protected while allowing the flow of health information needed to provide and promote high quality health care and to protect the public’s health and well being.” This would mean that once a patient shares their PGHD with a healthcare provider or entity covered by HIPAA, it becomes protected under HIPAA and the healthcare organization becomes responsible for protecting the patient’s information. Any violations of HIPAA can lead to costly fines.

The Centers for Medicare and Medicaid Services (CMS), through the Medicare Access and CHIP Reauthorization Act of 2015 (MACRA), requires that healthcare providers integrate PGHD into EHRs using Certified Electronic Health Record Technology (CEHRT).

“For 2020, eligible hospitals, critical access hospitals, and dual-eligible hospitals will have to report to CMS on health information exchange and provider to patient exchange, among other objectives.”

–CMS

The Food and Drug Administration (FDA) regulates certain devices and mobile applications and in September 2019, issued a Policy for Device Software Functionsand Mobile Medical Applications to clarify what is regulated. Healthcare stakeholders need to be aware of such policies to ensure they do not inadvertently violate FDA regulations.

There are also regulations that incentivize the use of PGHD and healthcare stakeholders can benefit from leveraging PGHD in their operations. The CMS operates the Promoting Interoperability Program (Meaningful Use) (previously the EHR Incentive Program) which pays eligible professionals, eligible hospitals, and critical access hospitals (CAHs) for meeting a set of standards for the use of CEHRT as part of their practices. Included in this are incentives for incorporating PGHD or data from a non-clinical setting into healthcare operations.

PGHD, spurred by rapidly evolving healthcare technology, is transforming the healthcare sector. There are many benefits to be realized from healthcare stakeholders working with patients to leverage this data, making it work for the patient, theprovider, and the health system. However, there are also challenges, especially regarding data accuracy, security, and privacy.  The large volumes of PGHD being collected by the myriad of devices and mHealth apps available, combined with the need to protect the privacy and security of the patient’s data, make it critical to develop a proper framework for any system being implemented to collect, share, and use PGHD. When developing such a framework, be sure to carry out a check of any regulations or requirement that may impact the collection, sharing, and use of PGHD to ensure compliance.

If you are seeking to incorporate healthcare technology into your organization or need help with leveraging PGHD, contact Acuma Health and we will work with you to find solutions that best suit you needs.

Patient Generated Health Data from Smart Watches Ensures Accurate Seizure Tracking and Reporting

Epilepsy is one of the most common neurological illnesses with estimates of approximately 3.4 million people in the US experiencing active epilepsy in 2015. Despite advancements in treatment options and optimal medication management, nearly one-third of patients with epilepsy continue to have seizures. Seizures can negatively impact the overall quality of a person’s life due to their unpredictable nature, occurring at anytime and anywhere. Healthcare technology  has come to the aid of people who have seizures with the development of seizure monitors or seizure alert devices that can detect the onset of a seizure and make an alert so that the individual suffering the seizure can be aided quickly. Good seizure monitors can also provide clinicians with detailed seizure data that can be used in the management of epilepsy.

There are many seizure alert devices on the market including seizure bed alarms or mattress sensors, seizure bracelets and smart watches, and camera/video/infrared devices. The report, “Seizure detection, seizure prediction, and closed-loop warning systems in epilepsy,” explored various seizure detection and prediction systems and noted that accelerometers such as smart watches, detect changes in velocity and direction and may serve to detect motor seizures such as tonic–clonic or myoclonic seizures. The authors found that a smart watch was able to detect 7 out of 8 tonic–clonic seizures in a pilot study. It further noted that the SmartWatch, manufactured by Smart Monitor Inc.:

·         utilized pattern recognition and feature analysis in its built-in seizure detection algorithm

·         can synchronize with a smartphone application via Bluetooth to transmit seizure data to the user’s mobile phone

·         the app can then contact caretakers to alert them of ongoing seizures

It is evident that the patient generated health data collected by smart watches can provide accurate tracking and reporting of seizures. Seizure monitor technology and seizure alert devices are very useful for patient engagement and can produce improved patient outcomes through use of the patient generated health data that they provide. The use of seizure monitors such as smart watches allows for early intervention in patients experiencing seizures, preventing injury, lessening the severity of the seizure, and potentially preventing sudden unexpected death in epilepsy (SUDEP). Seizure monitors also provide objective data that can be leveraged by healthcare providers to adjust therapy, allowing for better management of the patient with epilepsy and resulting in cost savings from reduced hospitalizations.

To find out more about how smart watches and patient generated health data can ensure better management of patient care and costs, download the Guide to Leveraging Healthcare Technology to Improve Management of High Risk Patients.