Introduction: What AI is and how it works
Simply put, artificial intelligence (AI) refers to a computer’s independent ability to solve problems for which they have not been specifically previously programmed. Algorithms in the machine learning driving the computer’s “thoughts” provide the conceptual background for the computer’s ability to process data input, followed by making decisions based on that information. The machine needs to:
- Accept input about a problem from its environment
- Create a list of potential actions
- Use probability and logic to select the most potentially successful activities
Machine learning happens when a program reabsorbs past experiences and uses this memory data for informing future actions, allowing it to give priority to choices that give successful results, further increasing the chances of arriving at the right answer.
Natural language processing (NLP), machine learning and Artificial Intelligence are rapidly becoming the foundations on which healthcare organizations utilize to stay ahead of the deluge of data created by the adoption of electronic health records (EHR) in keeping with the first rule for both humans and robots of doing no harm.
How AI is Transforming Healthcare
Accenture recently reported an analysis of near term value of AI healthcare applications to determine how the cost of implementing AI technology compares wit its potential impact and gains. The report found that AI applications could save the U.S. economy approximately $150 billion annually by 2026.
Here are a few areas in which AI is transforming the healthcare field, with the resulting improvements in outcomes and cost savings in fewer ER admissions with more accurate data and higher reimbursement. At the same time, we need to seriously consider the all sides and implications of AI, especially regarding patient information integrity, without risking a backlash in big data.
Oncology Applications and Radiomics
Improved tumor recognition in MRI diagnostics: following data input, two sets of MRIs may be given to a computer; one set showing various brain tumors and one which doesn’t. The computer breaks down the images into readable patterns, enabling it to differentiate between the patterns most likely to indicate a cancerous tumor and those of healthy patients.
By enabling the development of “virtual biopsies”, AI is advancing the new field of radiomics, in which image-based algorithms are harnessed to define the properties and phenotypes of tumors, enabling the identification of aggressiveness levels and targeted treatment of cancers.
Clinical Decision-making and Population Health the Latest Developments
Due to the huge increases in the volume of current medical information, supercomputers are necessary to even stay abreast of the latest practice and big data developments in predictive analytics, genomics, population health management and support of clinical decisions. As with the cancer tumor illustration, medical researchers can apply the same techniques to the enormous amount of data suddenly available, using state-of-the-art analysis to improve understanding and treatment of cancer and other diseases.
Programs such as “Watson” use their natural language and semantic processing abilities to function in clinical decision support within some of the country’s major healthcare organizations.
By taking in millions of pages of data, including the latest academic literature, the Watson system offers suggestions for decisions, along with confidence intervals measuring the applicability of the course of action. The higher the confidence number is for a diagnosis, treatment plan or drug, the more likely the system will recommend it as the way to go.
Improving Healthcare Access in Undeserved Regions
Many parts of the world suffer from extreme shortages of qualified clinical personnel, making the need for access to better diagnostics critical. Using apps and other tools available to low-resource providers, AI can screen chest x-rays for tuberculosis and other diseases, reducing the need for on-site radiologists.
Using miniaturized instruments and much smaller incisions, surgical teams use pre-op records and other data with real-time surgical metrics to improve outcomes. Because this enhances the physician’s precision, the results show a 21 percent reduction in post-op hospitalizations. The estimated cost savings is $40 billion.
Virtual Nursing Assistants (VNAs) and Remote Monitoring
VNAs offer potential benefits of patient access 24/7 to support and answers, such as about medications, with monitoring around the clock. This application could benefit elderly or frail patients in rural or underserved areas where reliable in-home caregivers may be difficult to find.
With an AI program taking over such tasks as sending prescription refill requests to pharmacies, turning lights off or on, sending monitoring data to their providers via home health devices or even ordering transportation to their next appointment, the continuum and quality of care need not be compromised. Implementing VNAs could reduce unnecessary hospital visits and other provider burdens, with a potential savings of $20 billion.
EHR workload reduction while maintaining data integrity
Emphasis on maintaining data integrity and privacy per HIPAA regulations and AHIMA’s information governance and oversight rules requires providers to have transparent, uniform and consistent procedures and policies in place for data management.
Reliability is of utmost importance in healthcare services delivery, with quality measures of data and other information built into processes and systems throughout all stages of the information cycle.
Programs such as “Nuance” and others employ machine learning to reduce reporting time while improving the documentation quality. Improved interface via AI using voice-recognition, predictive analytics and other tools and techniques creates more intuitive use as it quickly sorts through a provider’s virtual “in-basket.”
By automating and reducing workflow tasks for providers and staff¸ AI allows clinicians to better prioritize urgent matters while saving time, with estimated savings of $18 billion.
Machine Learning and Medical Billing Services
Companies such as M-Scribe utilize the latest technologies to stay current with the newest regulatory and technological billing and reporting developments. Since 2002, we have helped practices in specialties of all kinds and sizes identify areas needing improvement and showing them how to increase their reimbursement rates and bottom line. We have used new technologies to develop an advanced analytics model which make use of medical practice historical data to predict future revenue and help us better forecast monthly collections.
Contact us at 770-666-0470 or email for a confidential free analysis of your practice’s billing needs and projected revenue goals.