How Artificial Intelligence Can Be Used in Healthcare: Applications and Future

Artificial intelligence (AI) is a powerful and disruptive area of computer science with the potential to fundamentally transform the practice of medicine and the delivery of healthcare. Healthcare systems globally face significant challenges in achieving the ‘quadruple aim’: improving population health, enhancing patient experience, improving caregiver experience, and reducing rising costs. Ageing populations, chronic disease burdens, and increasing global healthcare costs necessitate innovation. The COVID-19 pandemic has accelerated the need to leverage data-driven insights for patient care, highlighting workforce shortages and access inequities. The increasing availability of multi-modal data (genomics, economic, demographic, clinical, phenotypic) coupled with technology innovations like mobile, IoT, computing power, and data security creates a convergence moment for healthcare and technology. Cloud computing, in particular, provides the necessary capacity for analyzing large datasets at speed and lower cost, enabling the transition of effective and safe AI systems into mainstream healthcare. Technology providers are increasingly partnering with healthcare organizations to drive AI-driven medical innovation. As Satya Nadella, CEO of Microsoft, noted, “AI is perhaps the most transformational technology of our time, and healthcare is perhaps AI’s most pressing application.” This article outlines recent breakthroughs in how Artificial Intelligence Can Be Used In Healthcare For various applications, describes a roadmap for building effective AI systems, and discusses the possible future direction of AI-augmented healthcare systems.

What is Artificial Intelligence?

Simply put, AI refers to the science and engineering of creating intelligent machines through algorithms or sets of rules that mimic human cognitive functions like learning and problem-solving. AI systems have the potential to anticipate and deal with issues proactively, operating intentionally, intelligently, and adaptively. A key strength of AI is its ability to learn and recognize patterns and relationships from large, multidimensional, and multimodal datasets. For example, AI systems could potentially translate a patient’s entire medical record into a single number representing a likely diagnosis. Moreover, AI systems are dynamic and autonomous, learning and adapting as more data becomes available.

AI is not a single universal technology but represents several subfields, such as machine learning (ML) and deep learning, which add intelligence to applications individually or in combination. Machine learning is the study of algorithms allowing computer programs to automatically improve through experience. ML can be categorized as ‘supervised’, ‘unsupervised’, and ‘reinforcement learning’ (RL), with ongoing research in sub-fields like ‘semi-supervised’, ‘self-supervised’, and ‘multi-instance’ ML. For those interested in formal training, a pg program in artificial intelligence and machine learning can provide the necessary expertise to work in this evolving field.

How to Build Effective and Trusted AI-Augmented Healthcare Systems?

Despite significant focus over the past decade, the adoption of AI in clinical practice remains limited, with many AI products for healthcare still in the design and development stage. Often, attempts are made to apply AI solutions to healthcare problems without considering the local context, such as clinical workflows, user needs, trust, safety, and ethical implications.

We believe AI amplifies and augments human intelligence rather than replacing it. Therefore, when building AI systems in healthcare, the focus should be on enhancing and improving the efficiency and effectiveness of human interaction, not replacing crucial elements. Innovations in healthcare AI will come from an in-depth, human-centered understanding of the complexity of patient journeys and care pathways.

A problem-driven, human-centered approach, adapted from various frameworks, is crucial for building effective and reliable AI-augmented healthcare systems. This involves multiple iterative steps.

Design and Develop

The first stage is to design and develop AI solutions for the right problems using a human-centered AI and experimentation approach, actively engaging appropriate stakeholders, especially the healthcare users themselves. Access to the necessary infrastructure and computing power is essential, reflecting the growing investment in areas like ai brain stock as the demand for powerful AI processing increases.

Stakeholder Engagement and Co-Creation

Build a multidisciplinary team including computer and social scientists, operational and research leadership, and clinical stakeholders (physicians, caregivers, patients), alongside subject experts (e.g., biomedical scientists). This team should include authorizers, motivators, financiers, conveners, connectors, implementers, and champions. A multi-stakeholder team provides the technical, strategic, and operational expertise needed to define problems, goals, success metrics, and intermediate milestones.

Human-Centered AI

A human-centered AI approach combines an ethnographic understanding of health systems with AI development. Through user-designed research, first understand the key problems, including needs, constraints, workflows, facilitators, and barriers to AI integration in clinical contexts. After defining problems, identify which are appropriate for AI and whether applicable datasets are available. By contextualizing algorithms within existing workflows, AI systems operate within norms, ensuring adoption and providing relevant solutions for the end user.

Experimentation

Focus on piloting new, stepwise experiments to build AI tools, using tight feedback loops from stakeholders for rapid experiential learning and incremental changes. Experimentation allows trying out new ideas simultaneously, exploring what works, what doesn’t, and why. This process helps clarify the purpose and intended uses for the AI system, the likely end users, and potential harm and ethical implications (e.g., data privacy, security, equity, safety).

Evaluate and Validate

Next, iteratively evaluate and validate the predictions made by the AI tool to test its functioning. Evaluation is based on three dimensions: statistical validity, clinical utility, and economic utility.

• Statistical validity: Understanding AI performance on metrics like accuracy, reliability, robustness, stability, and calibration. High model performance in retrospective, in silico settings is insufficient to demonstrate clinical utility.
• Clinical utility: Evaluating the algorithm in a real-time environment on hold-out and temporal validation sets (e.g., longitudinal and external geographic datasets) to demonstrate clinical effectiveness and generalizability.
• Economic utility: Quantifying the net benefit relative to the cost of investment in the AI system.

Scale and Diffuse

Many AI systems are initially designed to solve a problem at one healthcare system based on its specific patient population and context. Scaling AI systems requires special attention to deployment modalities, model updates, the regulatory system, variation between systems, and the reimbursement environment.

Monitor and Maintain

Even after clinical deployment, an AI system must be continually monitored and maintained to track risks and adverse events using effective post-market surveillance. Healthcare organizations, regulatory bodies, and AI developers should cooperate to collect and analyze relevant datasets for AI performance, clinical and safety risks, and adverse events.

What are the Current and Future Use Cases of AI in Healthcare?

Artificial Intelligence Can Be Used In Healthcare For numerous applications, enabling systems to achieve the ‘quadruple aim’ by democratizing and standardizing connected and AI-augmented care, precision diagnostics, precision therapeutics, and ultimately, precision medicine. Research into AI applications in healthcare is accelerating rapidly, with potential use cases across physical and mental health, including drug discovery, virtual clinical consultation, disease diagnosis, prognosis, medication management, and health monitoring.

Here is a non-exhaustive overview of AI applications in healthcare across the near term, medium term, and longer term, highlighting AI’s potential to augment, automate, and transform medicine.

AI Today (and in the Near Future)

Currently, AI systems do not reason like human physicians, who draw upon common sense or clinical intuition. Instead, AI resembles a signal translator, identifying patterns in datasets. Today, AI systems are being adopted by healthcare organizations to automate time-consuming, high-volume repetitive tasks. Considerable progress is also being made in demonstrating AI’s use in precision diagnostics (e.g., diabetic retinopathy screening and radiotherapy planning).

AI in the Medium Term (the Next 5–10 Years)

In the medium term, we anticipate significant progress in developing powerful, efficient algorithms (requiring less data for training) capable of using unlabeled data and combining disparate structured and unstructured data, including imaging, electronic health data, multi-omic, behavioral, and pharmacological data. Healthcare organizations and medical practices will evolve from AI platform adopters to co-innovators with technology partners in developing novel AI systems for precision therapeutics.

AI in the Long Term (>10 Years)

In the long term, AI systems will become more intelligent, enabling healthcare systems to achieve a state of precision medicine through AI-augmented and connected care. Healthcare will shift from traditional one-size-fits-all medicine to a preventative, personalized, data-driven disease management model, aiming for improved patient outcomes (better patient and clinical experiences) within a more cost-effective system.

Connected/Augmented Care

AI could significantly reduce healthcare inefficiencies, improve patient flow and experience, and enhance caregiver experience and patient safety throughout the care pathway. For instance, AI can be applied to remote patient monitoring (e.g., intelligent telehealth via wearables/sensors) to identify and provide timely care for patients at risk of deterioration.

In the long term, we expect healthcare clinics, hospitals, social care services, patients, and caregivers to be connected to a single, interoperable digital infrastructure using passive sensors combined with ambient intelligence. Two examples of AI applications in connected care are:

Virtual Assistants and AI Chatbots

AI chatbots (like Babylon and Ada) are used by patients to identify symptoms and recommend actions in community and primary care settings. They can integrate with wearable devices to provide insights to patients and caregivers on improving behavior, sleep, and general wellness.

Ambient and Intelligent Care

We also note the emergence of ambient sensing, which does not require peripherals.

Precision Diagnostics

Diagnostic Imaging

Automated classification of medical images is a leading AI application today. A review of AI/ML-based medical devices approved in the USA and Europe (2015–2020) found that over half (58% in the USA, 53% in Europe) were approved for radiological use. Studies demonstrate AI’s ability to meet or exceed human expert performance in image-based diagnoses across specialties like radiology (pneumonia detection), dermatology (skin lesion classification), pathology (lymph node metastases detection), and cardiology (heart attack diagnosis). This area benefits greatly from advancements in vision ai technologies tailored for medical imagery.

We expect widespread adoption and scaling of AI-based diagnostic imaging in the medium term. Two use cases include:

Diabetic Retinopathy Screening

Key to reducing preventable vision loss from diabetes is timely screening. However, screening is costly due to the large number of patients and limited eye care staff worldwide. Studies in the USA, Singapore, Thailand, and India show automated AI algorithms for diabetic retinopathy have robust diagnostic performance and cost-effectiveness. The FDA-approved AI algorithm ‘IDx-DR’ even received Medicare reimbursement for its use in detecting more-than-mild diabetic retinopathy with high sensitivity and specificity.

Improving Precision and Reducing Waiting Times for Radiotherapy Planning

An important AI application is assisting clinicians with image preparation and planning for cancer radiotherapy. Manual segmentation of images is time-consuming. The AI-based InnerEye open-source technology can reduce preparation time for head, neck, and prostate cancer by up to 90%, dramatically cutting waiting times for potentially life-saving treatment. Potential applications for the InnerEye deep learning toolkit include quantitative radiology for monitoring tumour progression, planning for surgery and radiotherapy planning.

InnerEye deep learning toolkit applications for quantitative radiology, tumor monitoring, surgery planning, and radiotherapy planning in healthcare

Precision Therapeutics

To achieve precision therapeutics, we need to improve our understanding of disease significantly. Researchers collect multimodal datasets exploring cellular and molecular disease bases, leading to digital and biological biomarkers for diagnosis, severity, and progression. Two important future AI applications include immunomics/synthetic biology and drug discovery.

Immunomics and Synthetic Biology

Applying AI tools to multimodal datasets may allow us to better understand the cellular basis of disease and patient clustering, leading to more targeted preventive strategies, such as using immunomics for diagnosis and treatment prediction. This could revolutionize care standards, particularly for cancer, neurological, and rare diseases, personalizing the patient experience.

AI-Driven Drug Discovery

AI will significantly improve clinical trial design, optimize drug manufacturing, and replace combinatorial optimization processes in healthcare. Initiatives like DeepMind’s AlphaFold, which predicts protein structures, set the stage for better disease understanding, predicting protein structures, and developing more targeted therapeutics for both rare and common diseases.

Precision Medicine

New Curative Therapies

Synthetic biology has produced developments like CRISPR gene editing and personalized cancer therapies. However, developing these advanced therapies remains inefficient and expensive.

In the future, with improved access to multi-omic data (genomic, proteomic, glycomic, metabolomic, bioinformatic), AI will handle greater systemic complexity, transforming disease understanding, discovery, and biological manipulation. This will improve drug discovery efficiency by better predicting effective agents and anticipating adverse effects earlier, potentially democratizing access to novel therapies at lower costs.

AI-Empowered Healthcare Professionals

In the longer term, healthcare professionals will leverage AI to augment care delivery, providing safer, standardized, and more effective care. Clinicians could use an ‘AI digital consult’ to examine ‘digital twin’ models of patients (a truly digital and biomedical version) to test the effectiveness, safety, and experience of interventions (like cancer drugs) in a digital environment before real-world application.

Challenges

Significant challenges exist related to the wider adoption and deployment of AI into healthcare systems. These include data quality and access issues, technical infrastructure limitations, organizational capacity constraints, and the critical need for ethical and responsible practices, alongside safety and regulation aspects. While some have been discussed, others extend beyond the scope of this article. Addressing potential biases, like racist ai examples that can arise from biased training data, is paramount for equitable healthcare AI.

Conclusion and Key Recommendations

Advances in artificial intelligence can be used in healthcare for transforming many aspects, enabling a more personalized, precise, predictive, and portable future. Whether adoption is incremental or radical, the impact of these technologies demands health systems adapt. For systems like the NHS, AI has the potential to free up healthcare professionals’ time, allowing focus on patient needs, leveraging global data assets and advanced knowledge to deliver high-standard care universally. Globally, AI could become a key tool for improving health equity.

While the last decade focused on digitizing health records for efficiency and billing, the next decade will focus on gaining insight and value from these digital assets, translating them into improved clinical outcomes with AI assistance, and creating new data assets and tools. We are at a turning point in the convergence of medicine and technology. Despite opportunities, formidable challenges exist regarding real-world implementation and scale. Expanding translational research in healthcare AI applications is key to realizing this vision. Alongside this, investment is needed to upskill the healthcare workforce and future leaders to be digitally enabled, embracing rather than being intimidated by AI-augmented healthcare. The landscape of new ai developments requires continuous learning and adaptation.

Healthcare leaders planning to leverage AI for health should consider these minimum issues:

• Processes for ethical and responsible data access: healthcare data is highly sensitive, often inconsistent, siloed, and not optimized for ML development, evaluation, and adoption.
• Access to domain expertise/prior knowledge to interpret datasets and create necessary rules for generating insight.
• Access to sufficient computing power for real-time decisions, exponentially enhanced by cloud computing.
• Implementation research: critically, exploring and researching issues arising when deploying algorithms in the real world to build ‘trusted’ AI embedded in appropriate workflows.

References

1
2
3
4
5
6
7
8
9
10
11
12
13
14
19
20
21
22
23
24
25
26
27
28
29
30
31
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50