Overseas Viewpoint
The number of artificial intelligence tools in the healthcare field is growing, and they are expected to help the medical system eliminate many factors that cause medical inequity and help achieve medical equality. On August 29, Harvard Business Review published an article by Carol Cruickshank, partner and chairman of Kearney Americas, Cian Wade, MD, and Junaid Bajwa, chief medical scientist at Microsoft, introducing some new AI tools in the medical field and explaining why they have great potential in solving seemingly insurmountable medical challenges.
In the medical field, artificial intelligence is both exciting and unsettling. Optimists see clear potential to reform the efficiency and quality of health care, while pessimists worry that these tools will exacerbate society's already significant health inequalities by prioritizing the wealthiest and healthiest populations.
Are these concerns justified? Will the new tools we use only widen inequality gaps in health outcomes? While AI should be applied with caution, we are confident in its vision as a democratizing force that health care desperately needs. Imagine the journey of a patient with a long-term illness. At every stage, countless factors influence their health outcomes: the patient’s language and literacy skills, their ability and willingness to navigate complex health systems, and the biases of the health care providers and medical knowledge base used to treat their condition. This is where AI comes into play as a revolutionary tool that can enable the health system to provide better care for everyone, especially the most vulnerable and underserved populations. AI’s ability to leverage many different types of data to predict and intervene at every stage of a patient’s treatment journey puts it in a unique position to address health inequities .
Identifying High-Risk Patients
Preventive care is the most cost-effective and effective clinical intervention point to improve patient health, which means engaging patients in healthy behaviors, vaccinations, and managing risk factors such as hypertension and obesity. However, vulnerable patient populations often lack awareness of the steps they should take. It is critical for healthcare providers to identify and proactively reach out to patients who may be at risk for adverse health outcomes. A patient's health record may have subtle signs that they are at risk, such as a missed appointment or a short note from a previous consultation. But it is extremely challenging to identify these markers and take action. Artificial intelligence can understand this highly complex patient data and provide personalized services for immunizations, disease screening, and advance care planning, thereby reducing hospitalization rates and costs.
Overcoming Communication Barriers
Once a patient enters the healthcare system for an assessment of their problem, it is critical that they communicate effectively with their care team. For patients with language barriers, low health literacy, or neurocognitive conditions, communication gaps put them at higher risk for poor health outcomes. AI opens the door to better communication by serving as a liaison between the healthcare system and patients, providing more understandable information about risks, symptoms, and treatment plans. For example, one company has used generative AI to create a multilingual and compassionate digital front desk for patients. This omnichannel platform helps patients schedule care, collect relevant admission information, manage the billing process, and send personalized reminders in the language and communication medium that best suits the patient.
These solutions are also critical to improving productivity in a healthcare system that is increasingly stretched to its limits. By streamlining and automating administrative tasks associated with patient access and assessment, healthcare professionals can refocus their time on clinical work, delivering the greatest value in improving health outcomes for the most medically complex patients.
Solve The Problem Of Resource Acquisition Differences
The “law of reverse care” states that patients with the highest needs often have the least access to health resources. Many patients in rural areas are far from professional medical services or may lack social networks that are critical to their health and recovery, and artificial intelligence has shown initial signs of helping policymakers overcome these barriers and provide important solutions. For example, companies have developed AI-driven smartphone applications that can assess symptoms, diagnose a variety of medical problems, and make personalized care recommendations. The application has even surpassed human doctors in accurately diagnosing the causes of rheumatism, rashes, and abdominal pain during emergency room visits. These AI-based tools provide patients with highly effective and scalable “pocket doctors” who can access services in the most appropriate way, no matter how far away they are from their healthcare providers.
AI can also improve outcomes after treatment. Currently, patients who experience healthcare inequities are at higher risk for deterioration and hospital readmission, but AI can fill the gap by enabling these patients to understand the signs of clinical deterioration and when and how to access urgent medical care.
Reducing The Impact Of Human Bias
There is a large and worrying body of evidence suggesting that healthcare professionals may be perpetuating inequities by failing to recognise the severity of illness due to unconscious bias, misattributing symptoms to cultural stereotypes. A relevant example is access to appropriate pain relief. When black patients present with abdominal pain in a similar manner, they are only half as likely to receive opioid treatment as white patients. Artificial intelligence may be an important tool in counteracting the effects of these human biases. Recent research has shown that AI can be highly effective in detecting pain in patients, with a model trained on over 140,000 facial images of patients in pain and those without pain, and the assessments were 88% consistent with those of expert doctors, demonstrating that it can be a scalable solution for rapid pain assessment.
Bias can also lead to misdiagnosis of important conditions and affect treatment decisions. In one study, black and Asian patients whose echocardiograms showed signs of aortic stenosis had a 22% and 25% lower diagnosis rate than white patients. An AI tool designed to improve echocardiogram diagnostic performance has been piloted by a company and has great potential to reduce such inequities. It uses natural language processing capabilities to process echocardiogram reports , improving the diagnosis rate of aortic valve stenosis by 35% compared to a physician's interpretation of the same report, allowing patients to be provided with appropriate treatment quickly.
Enhancing Diversity In Clinical Trials
Lack of diversity in clinical trials is one of the key causes of health inequities. For example, when albuterol, the world’s most prescribed inhaled bronchodilator , came to market, about 95% of lung disease studies were conducted in people of European ancestry. This narrow testing did not indicate the drug’s differential effectiveness in genetically diverse populations. African American children are known to respond less well to albuterol compared with their white peers. This lack of diversity in testing data may partially explain the threefold higher death rate from asthma-related causes in this group.
Pharmaceutical companies are now using AI to recruit more diverse and representative patient populations. Experiments have shown that relaxing the eligibility criteria of the original trial can increase the diversity of the study population without affecting safety or effectiveness. In addition, AI-based analysis can also enable researchers to reach and recruit a wider and more diverse range of patients to participate in clinical trials.
Putting Fairness At The Heart Of AI Growth
The barriers to entry for new AI startups and products are falling, and a steady stream of funding is driving explosive growth in AI. But ensuring that these innovations benefit patients facing healthcare inequities requires all parties to prioritize this goal.
Wellth , an AI-enabled digital health platform , uses behavioral economics incentives to improve patient adherence to treatment plans, especially among those receiving Medicaid, a group of patients who experience healthcare inequities more often and have historically been unattractive from an investment perspective due to low reimbursement rates. Patients using the Wellth platform experienced an average 42% reduction in hospital utilization and a 16% improvement in medication adherence (including non-Medicaid patients).
Progress in implementing AI for vulnerable patient populations is being slowed by stakeholder concerns that a lack of diversity in training data could lead to models being biased against such groups. To combat potential bias, healthcare companies are validating the performance and fairness of clinical algorithms by helping AI developers assess bias across geographies and demographics.
Aligning Financial Incentives With Fairness
Another way to invest in AI-based innovations that could reduce inequalities is through the adoption of alternative payment models ( APMs), which provide health systems with budgets to manage the care cycle for patients or populations. APMs involve the sharing of financial risk between health systems and payers. These models encourage health systems to invest in preventive care to avoid higher subsequent costs, to coordinate effectively between multidisciplinary teams managing complex conditions, and to provide ongoing support to patients to avoid treatment failure. Therefore, health systems that contract on an APM basis will see value in investing in AI-based tools that help carry out activities that reduce health inequalities .
Intergenerational Opportunities
AI has the potential to improve health outcomes at every point in the patient journey and address many of the root causes of health inequities . Turning this into a scalable reality will require significant changes in financing models and the strategic priorities of health systems. Thoughtful approaches must be developed to ensure that tools are designed with equity in mind and that AI innovations do not leave vulnerable patients behind. As health systems around the world grapple with the ethical and financial costs of health inequities , the case for using AI to close gaps couldn’t be more compelling. Finally, it’s important to remember that AI tools that improve the quality of care for the most vulnerable patients will often also improve care for everyone.
No comments:
Post a Comment