INTRODUCTION
AI-assisted predictive diagnoses for better healthcare
Using AI to strengthen preventative medical care stops unnecessary diagnostic tests, incorrect diagnoses, and unwanted procedures. Engaging patients in their medical care dramatically reduces the likelihood of being readmitted for easily treatable and preventable conditions. Thus, patient-centered approaches drive measurable cost savings across community healthcare systems.
The prevalence of chronic diseases is dramatically increasing and impacting mortality worldwide. In the United States, $3.7 trillion is spent annually on addressing chronic disease impact—approximately 19.6% of the gross domestic product!
However, most chronic disease pathologies are caused by just a few key risk factors:
- Tobacco use and exposure to secondhand smoke.
- Poor nutrition, including diets low in fruits and vegetables and high in sodium and saturated fats.
- Physical inactivity.
- Excessive alcohol use.
As healthcare organizations continue to embrace person-centered care initiatives, there is tremendous potential to reduce the societal impacts caused by chronic disease. Soon, physicians will be better equipped with laser-focused diagnostic insights on patient physiology, risk factors, and individual treatment plans.
Medical research focused on early diagnosis utilizes artificial intelligence (AI), machine learning (ML), and artificial neural networks (ANN) with particle swarm optimization (PSO).
AI-based predictive analytics is on the front-lines of innovation to better predict diseases by studying data generated from health devices, sensors, clinical databases, social networks, wearables, and the medical Internet of Things (MIoT).
BREAKTHROUGHS
Cutting-edge AI models developed to improve healthcare
Medical researchers at the Perelman School of Medicine at the University of Pennsylvania and the University of Florida College of Medicine recently received a $4.7 million grant from the National Institutes of Health (NIH) to apply AI and ML algorithms to information from patient medical records to better predict the risk of rare disease development.
Over the next four years, the team will continue developing predictive models to help understand which patients are more likely to experience five types of vasculitis and two types of spondyloarthritis.
The predictive model being developed through this grant is called “PANDA: Predictive Analytics via Networked Distributed Algorithms for multi-system diseases” and was developed by principal investigators Yong Chen, Ph.D., a professor of Bio-statistics, and Peter A. Merkel, MD, MPH, chief of Rheumatology and a professor of Medicine and Epidemiology at Penn, and Jiang Bian, Ph.D., chief data scientist of the University of Florida Health System and a professor in the Health Outcomes & Biomedical Informatics at the University of Florida College of Medicine.
“This is an exciting step forward, building on our current privacy-preserving distributed algorithms or PDA framework, from clinical evidence generation to AI-informed clinical decision-making interventions. Despite the clear need to reduce the dangerous and costly delays in diagnosis, individual clinicians, especially in primary care, face important challenges. The proposed machine learning algorithms will adaptively update their key parameters as more data are available,” said Chen. “We plan to evaluate these machine learning algorithms periodically to ensure they meet our pre-specified standards and can evolve positively over time.”
Soon, artificial intelligence and machine learning will continue to transform the ability of medical clinicians to make more accurate, data-informed decisions. Smart AI and machine learning-based solutions are the future of preventive medicine.
USES
AI support for healthcare comes in different forms
Artificial intelligence-powered healthcare technologies are speeding up core medical diagnostic processes and even delivering higher accuracy than many human clinicians are able to achieve. How do AI interventions help?
- Machine learning solutions can rapidly compare millions of different pieces of medical data to quickly and reliably enhance CT imaging to deliver better cancer screening processes.
- Workflow process automation leveraging artificial intelligence paired with medical robotics augment traditional medical interventions in patient care areas such as daily hygiene, surgical prep, remote patient monitoring, and other vital areas.
- Automation handles repetitive administrative tasks, freeing up clinicians to spend more time with patients.
- With predictive analytics,the ability to identify, limit, and even prevent severe pathologies from expanding into chronic diseases is greatly enhanced.
- AI-ML is being used in many web and mobile applications to build conversational interfaces that better connect patients with the services in their community health settings.
- As telehealth continues to expand, AI-enabled automation enables providers to more reliably serve patients through virtual and in-person care visits.
- Chatbot services provide patients access to life-saving medical insights, guidance, and the ability to diagnose everyday community health issues remotely and reliably. These new AI-powered chatbot systems can connect patients directly to pharmacy services after successfully diagnosing a treatable medical problem.
METHOD
How does AI disease detection work?
On October 10th, 2022, Mahmood Lab of Harvard Medical School published a paper in Nature Biomedical Engineering outlining a novel method for using A.I. to detect diseases called SISH (Self-Supervised Image Search for Histology).
“We show that our system can assist with the diagnosis of rare diseases and find cases with similar morphologic patterns without the need for manual annotations and large datasets for supervised training,” said senior author Faisal Mahmood, assistant professor of pathology at HMS at Brigham and Women’s. “This system can potentially improve pathology training, disease subtyping, tumor identification, and rare morphology identification.”
SISH works similarly to how a search engine works by analyzing different diagnostic images of patients and comparing them to large datasets of disease pathologies and treatments. This application uses deep learning to constantly gain new insights and learn from the ever-expanding medical records database.
The SHISH application studied by Harvard used 22,385 diagnostic whole-slide images across 13 anatomic sites and 56 disease subtypes. The researchers were able to identify limits but found it delivered reliable and scalable results that vastly improved the detection of many complex disease subtypes such as heart disease, kidney disease, and cancers.
In this application, AI detects diseases by comparing images of patients against records of their medical diagnoses.
FUTURE
AI-powered predictive analytics to take center stage
According to the Centers for Disease Control (CDC), six in 10 U.S. adults have at least one chronic disease, and four out of 10 with two or more. By 2025, more than half of the U.S. population will suffer from chronic disease pathologies connected to heart disease, cancer, and diabetes. The application of preventive, predictive, and prescriptive analysis will help to interpret individual and community health and take the correct action.
Predictive analytical models reveal if a chronic disease can develop
Preventive analytics
Preventative medicine is a focal point of value-based care initiatives and person-centered care approaches. Preventive models help patients avoid the onset of debilitating disease pathologies before they develop into severe or more chronic syndromes.
Preventive data analytics explores large data sets from descriptive and diagnostic analytics to identify and correct disease pathologies before they become chronic.
Predictive analytics
Predictive models uncover trends from past health data of patients to impact and shape future outcomes. AI and ML learning models can dive deep into patient records and compare them to family records or insights from the local community and patients with similar disease pathologies, physiology, or many other factors. Predictive trends reveal risk areas where chronic disease pathologies could develop if left unchecked.
Predictive analytics provides healthcare teams with additional insights into how symptoms are driving disease and which interventions are most likely to be effective according to evidence-based insights.
Prescriptive analytics
Prescriptive human intelligence can combine unstructured datasets to look at broader cultural and socio-economic trends and the environmental factors that drive individual and family health.
Prescriptive analytics offer insights into how personal health information matches up with macro-level societal, demographic, environmental, and other social determinants of health.
Thus, AI and ML models provide data scientists with many methods to compare various forms of information to drive medical decision-making that deters the spread of chronic diseases.
