Keyword

1 Introduction

As a new chapter unfolds in healthcare, ubiquitous AI shines as a light for low and middle-income countries, turning distant aspirations into achievable outcomes.

Artificial Intelligence (AI) is a robust developing branch of computer science designed to build machines that can mimic human beings and perform tasks equally or better (Shimizu and Nakayama 2020). The widespread adoption of AI technologies into our daily lives has become ubiquitous, making everyday items and settings, such as cellphones, wearable technology, and homes, smarter and more interactive. Over the past decade, the world has witnessed how AI revolutionized various industries, including healthcare, with machine learning and deep learning techniques enabling the development of medical devices that can aid in diagnosing, screening, predicting, and treating diseases (Esteva et al. 2017). In this era of big data, AI devices can assist researchers and physicians in analyzing large amounts of data and identifying possible connections leading to new insights and discoveries (Obermeyer and Emanuel 2016). In fact, up to 520 marker-cleared AI medical algorithms have already received FDA approval, with more expected to follow (FDA 2022). Developed countries have already begun reaping the benefits of advanced AI techniques in daily clinics. Despite limited access nowadays, this trend into shifting to digital healthcare could be especially noteworthy in low and middle-income countries (LMICs), where traditional healthcare systems often struggle to meet the population’s needs.

This chapter will explore the challenges and opportunities of adopting AI in healthcare. We will discuss potential applications of ubiquitous AI in LMICs, including medical imaging, diagnostic decision support, telemedicine, and electronic health records. Despite the challenges ahead, the increasing availability and adoption of AI technology suggest that its impact on the future of healthcare is both exciting and unique.

2 Potential Applications of Ubiquitous AI in LMIC Healthcare

Having limited resources and infrastructure, low- and middle-income countries (LMICs) often struggle to provide their populations with the best available and the most quality-assured healthcare. While challenges exist, the adoption of AI can address the shortage of healthcare professionals by automating routine tasks, enabling early, timely detection and accurate diagnosis of diseases, tailoring treatment plans, and providing real-time monitoring of patients. The future of AI in the digitalization of medicine in LMICs is promising, and its potential must be harnessed.

The potential applications of ubiquitous AI in healthcare in LMICs are vast, ranging from disease diagnosis and personalized treatment plans to drug discovery and development. Improved health outcomes, reduced mortality rates, and enhanced training opportunities for healthcare professionals could be achieved by AI-powered tools with better management and analysis of medical data and increased access to healthcare through telemedicine and remote monitoring. Additionally, AI can improve public health surveillance and disease control while providing increased research opportunities for identifying and addressing healthcare disparities. Finally, the development and application of AI technology can improve resource management, ensuring that medical equipment, supplies, and staff are utilized efficiently and effectively, hence reducing the financial burden, and boosting the economy of LMICs.

One example can be using deep learning techniques for more accurate diagnoses. This can provide a more precise diagnosis based on data experience. The histopathologic slice from a country with limited resources and experience can be reviewed and verified within minutes, compatible with those of the experienced center with an expert histopathologist.

Other areas where AI can significantly impact include:

  • Medical image analysis.

  • Real-time monitoring of patients.

  • Predictive analytics for healthcare management.

We can envision this possible application in the example of Armenia. In several border-closed regions of Armenia, there are few specialized oncology radiologists, and patients must travel to the capital even for a cancer screening. If an AI-based tool were available to identify, for example, suspicious breast masses, a regional radiologist could spare patients without such lesions the need to travel to centralized hospitals for screenings. This would be a cost-effective approach and ensure timely detection and treatment for at-risk people.

Electronic health record (EHR) management and analysis is another area where ubiquitous AI could significantly impact LMICs. AI could automate data analysis and entry, allowing healthcare professionals to enter patient information and obtain relevant data swiftly and simply. This can improve the accuracy and completeness of medical records, serving as a more precise ground for future analyzes. Additionally, AI-powered analytics could identify patterns and trends in EHR data, allowing for more effective disease surveillance, outbreak detection, and targeted interventions. By implementing AI to improve EHR management and analysis, LMICs could significantly improve the quality and accessibility of healthcare services.

Ensuring that medical equipment and supplies are accessible when and where necessary is one of the biggest obstacles in healthcare delivery in LMICs. AI can play a critical role in addressing this challenge by providing advanced demand forecasting and inventory management tools. For instance, in Nigeria, the non-profit startup LifeBank has partnered to use AI-powered tools to optimize the supply chain for blood and other critical medical supplies. The demand, transport times, and storage capacity are the main factors to be analyzed by AI to ensure that medical supplies are available when and where they are needed. This has helped reduce waste and improve overall efficiency, enabling healthcare providers to deliver lifesaving care to more patients. By leveraging AI to improve supply chain management, healthcare providers in LMICs can overcome logistical challenges and improve access to essential medical supplies and equipment.

Another example of ubiquitous AI is wearable technology, which has great potential to be implemented in LMIC, where doctor visits tend to be infrequent and inconsistent. n such regions, wearables can empower patients to remotely monitor their health and well-being more efficiently, reducing the need for some in-person consultations. For instance, these devices can continuously track blood pressure, heart rate, and blood glucose levels, offering valuable insights for patients with chronic conditions like diabetes and delivering more precise information for physicians (Gao et al. 2016). This remote monitoring allows healthcare professionals to intervene promptly if any issues arise, enabling them to adapt and guide treatment accordingly.

Furthermore, wearables can be employed to monitor physical activity levels, sleep patterns, and other health-related metrics, giving both patients and doctors a holistic view of their health status. Leveraging AI algorithms to analyze the data gathered by these devices, clinicians and healthcare providers can uncover valuable insights into patient health, recognize trends and patterns in health data, and offer more tailored recommendations for patients.

2.1 AI-Driven Drug Development and Clinical Trials

Affecting various fields of healthcare, drug development is another aspect that is highly impacted by the implementation of AI. Particularly in resource-limited settings, these AI tools have the potential to shift the way new therapies are discovered and brought to market (Lecun et al. 2015). One of the most significant advantages of using AI in drug development over human resources is its ability to analyze vast amounts of data, such as genomic, proteomic, and clinical trial data, during a time that would be impossible for researchers. This enables AI algorithms to identify potential drug candidates and therapeutic targets more quickly and efficiently. An example of AI’s impact is its application in drug repurposing, where algorithms analyze existing drugs and their interactions with disease pathways to uncover new therapeutic uses (Ahmed et al. 2022). This approach expedites the process of drug discovery and development, as repurposed drugs have already undergone extensive safety testing and can often bypass early-stage clinical trials.

AI can also improve the efficiency of preclinical drug development, optimizing the design of trials and predicting the likelihood of success for specific drug candidates, ultimately conserving resources, and enabling the development of more affordable treatments for patients in LMICs (Vamathevan et al. 2019). By analyzing large datasets, AI algorithms can help researchers identify the most appropriate patient populations for clinical trials and forecast potential trial outcomes (Weissler et al. 2021). This can lead to more efficient trial designs and increased success rates, ultimately accelerating the development of new therapies. In LMICs, where patient recruitment for clinical trials may be challenging due to limited infrastructure and resources, AI can streamline the process and improve trial outcomes.

From the point of view of LMICs, AI-driven drug development can be particularly valuable in the context of diseases that are prevalent in these regions but may not have been of interest to the industries in higher-income countries to be addressed. For example, AI could identify new therapeutic targets and drug candidates for neglected tropical diseases, which disproportionately affect populations in LMICs. By using AI to develop new treatments for these diseases, researchers can help to reduce health disparities in LMICs. By incorporating AI-driven innovations into the drug development process, researchers can help to address many of the unique challenges faced by LMICs in healthcare and contribute to forming a healthier population.

2.2 AI-Powered Precision Medicine

In the context of LMICs, there is a substantial anticipated benefit for AI-powered precision medicine. Tailored treatment plans based on improved patient stratification can help address unique healthcare challenges in these countries, including limited resources and diverse patient populations facing a high burden of both communicable and non-communicable diseases.

AI algorithms can analyze large amounts of patient data, such as demographic, genetic, and clinical information, and detect meaningful patterns and correlations, which may elude human researchers (Obermeyer and Emanuel 2016). By harnessing this data-driven insight, healthcare providers in LMICs can develop more accurate diagnoses and better understand the factors contributing to specific diseases or conditions, ultimately leading to more targeted and effective treatments.

One example of AI-powered precision medicine in an LMIC context is using AI algorithms to predict the risk of tuberculosis (TB) among patients in high-burden settings. In countries where TB is a significant public health concern, like India, AI can help to identify patients who are most likely to develop active TB. By identifying high-risk patients with the factors such as genetic makeup, environmental exposure, and other health conditions, healthcare providers can target interventions and treatment plans more effectively, ultimately reducing the overall burden of TB in the population (Orjuela-Cañón et al. 2022; Schwalbe and Wahl 2020).

Personalized treatment plans can also help to optimize the use of limited healthcare resources in LMICs. Healthcare professionals can avoid using needless or ineffective tactics by tailoring therapies to each patient’s requirements, sparing those who might not benefit from unnecessary procedures. This is particularly important in LMICs, where healthcare budgets may be limited, and the need to allocate resources efficiently is paramount.

Furthermore, AI-driven precision medicine can help to address the unique challenges faced by LMICs in managing both communicable and non-communicable diseases. AI algorithms, for instance, can be used to identify possible interactions between non-communicable illnesses like diabetes and infectious diseases like HIV and create personalized treatment plans that consider these complications. This can lead to better patient outcomes and a more holistic approach to healthcare in LMICs. The potential for better patient treatment and outcomes in LMICs will only increase as AI is constantly evolving and is more widely integrated into global healthcare.

3 Successful Implementations of Ubiquitous AI in LMICs Healthcare

Besides the potential that can be brought, the benefits of AI tools in LMICs are becoming increasingly evident, with some countries already experiencing successful implementations. One such example is Rwanda, where the government is utilizing AI-powered drones to deliver medical supplies, including blood units, to remote areas of the country. Advanced sensors on the drones are capable of detecting fluctuations in temperature and humidity, precisely monitoring vaccine and medical supply transportation, to ensure that supplies are kept at the appropriate temperature throughout the delivery process. With 14 drones currently serving 21 hospitals, the program has already delivered over 20,000 blood units, helping to save countless lives (The Guardian 2022). This cutting-edge application of AI in healthcare demonstrates how technology has the power to close gaps in healthcare delivery and allocate resources where needed the most.

Another successful application of AI tools is the use of telemedicine. The use of modern communication technologies, such as video consultations and conferencing, to discuss complex patient cases with external experts is referred to as telemedicine. One example of successful telemedicine implementation is in the Pediatric Cancer and Blood Disorders Center of Armenia, which established four multidisciplinary working groups to discuss all cases of soft tissue and bone tumours during weekly meetings with an expert from the University Hospital of Münster (UKM) in Germany using VITU (Virtuelles Tumorboard) software. This telemedicine platform is an example of ubiquitous technological application in LMICs, as it provides a unique opportunity for specialists from developing countries to establish effective communication with international experts, increasing educational, practical, and scientific opportunities for local healthcare providers, and improving outcomes of pediatric cancer care (Hovhannisyan et al. 2020).

AI-powered chatbots are among the most popular forms of ubiquitous AI. One such chatbot is “Sehat Kahani,” which operates in Pakistan and aims to help connect patients in remote areas with healthcare providers (Khan 2023). Using natural language processing (NLP), the chatbot understands the patients’ symptoms and provides medical advice and referrals to healthcare facilities. Other examples of ubiquitous AI chatbots in healthcare for LMICs include mDiabetes in India, which aims to help people with diabetes manage their condition using SMS messaging, Ada in Ethiopia, which uses AI to diagnose and treat common illnesses; and SARA in South Africa, which helps HIV-positive patients manage their condition by providing personalized information and reminders.

Despite the various areas in which AI tools are implemented, diagnostics remains one of the most prominent. For example, in India, as well as in Nigeria, Ghana and South Africa, researchers are using AI to screen for TB. By analyzing chest X-rays, the AI tool can detect signs of TB and enable healthcare workers to intervene earlier, improving patient outcomes (Khan et al. 2020).

These are just a few examples of successful AI tool implementations in LMICs demonstrating how innovative technologies are revolutionizing healthcare by improving outcomes and delivering healthcare resources to remote areas. With continued advancements in AI, we can expect to see even more exciting innovations in healthcare be approved worldwide.

4 Diabetic Retinopathy Screening in India: A Case Study

India’s population of over 1.3 billion people presents a unique challenge for healthcare, with a shortage of doctors being one of the most pressing issues. However, the Indian government has taken steps to integrate AI into the healthcare system, investing in several initiatives such as the National Institution for Transforming India (NITI) Aayog national program on AI, National Health Stack, and collaborations with companies like IBM Watson Health, Microsoft Healthcare, and GE Healthcare (NITI Aayog 2018). A prime example of successful implementation is the use of AI for diabetic retinopathy (DR) screening (Ting et al. 2016). DR is a complication affecting one in three people with diabetes and can cause blindness if left untreated. India has more than 72 million people with diabetes, and DR prevalence is estimated at around 18% (Brar et al. 2022).

AI has been used to diagnose DR accurately, efficiently, and cost-effectively by analyzing fundus images and extracting clinically relevant information from medical data. AI systems have been implemented by organizations like Aravind Eye Care System (AECS), the Indian Institute of Technology (IIT) Madras, and Sankara Nethralaya (Raman et al. 2019). These systems have achieved accuracy rates of over 85–95% in detecting DR, reducing the workload of ophthalmologists, and making DR screening more accessible and affordable.

The implementation of AI in DR screening has had a significant impact on the Indian healthcare system. Increased accessibility, affordability, and efficiency have led to earlier detection and treatment of DR. Reduced workload for ophthalmologists allows them to focus on more complex cases, and the overall reduction in screening costs makes DR screening more affordable for low-income individuals. However, addressing the ethical, social, economic, and legal consequences of AI applications in DR screening remains an important consideration (Rajalakshmi 2019; Rajalakshmi et al. 2018).

5 Challenges and Opportunities for AI in LMIC Healthcare

Despite potential benefits that could be anticipated, there are a few significant limitations to the implementation of ubiquitous AI in healthcare in LMIC that need to be addressed.

5.1 Challenge 1: Price

One of the biggest challenges is the limited funding available for the procurement and maintenance of AI technology in healthcare facilities. The high cost of purchasing and maintaining AI systems can be prohibitively expensive, limiting its wide adoption. For example, while conventional MRI machine can cost around $500,000 and $1m USD, an MRI with AI-power can be between $1.5m and $3m USD, with additional expenses for installation, maintenance, and staff training. Other AI-enabled healthcare systems, like software for predictive analytics and electronic health data, can also be quite expensive hindering their adoption (Strubell et al. 2019).

These high costs can be especially prohibitive for healthcare facilities in LMIC, where budgets for healthcare are already limited. In comparison to high-income countries, where healthcare budgets can be in the billions of dollars. This means that healthcare facilities in LMIC may have to make difficult choices about where to allocate their limited resources, purchasing an AI-enabled MRI machine or investing in other critical healthcare infrastructure, such as hiring additional healthcare professionals or building new medical facilities (Wilson and Khansa 2018).

In contrast, high-income countries make huge investments, for instance, the government of the United States has allotted billions of dollars in financing for the creation and application of AI technology in healthcare through programs like the Precision Medicine Initiative and the All of Us Research Program of the National Institutes of Health (Sankar and Parker 2017). The financial means to engage in AI technology are also available to private healthcare organizations, some of which have even established their own AI research laboratories to create and test novel AI-enabled healthcare systems.

5.2 Ways to Overcome Challenge 1: Price

To address the issue of limited funding for AI technology in healthcare in LMICs, there have been efforts to develop more affordable and accessible AI systems. For example, some companies have developed cloud-based AI systems that can be accessed remotely by healthcare facilities, reducing the need for expensive hardware and infrastructure (Mollura et al. 2020). Additionally, there have been calls for increased international funding and collaboration to support the development and implementation of AI technology in healthcare in LMICs.Another solution to overcome the high cost of AI technology in healthcare is through public-private partnerships. In some cases, private companies may be willing to invest in the implementation of AI technology in healthcare facilities in LMICs, for example in exchange for access to medical data. This can provide a mutually beneficial partnership that enables healthcare facilities to adopt AI technology at a lower cost, while also providing private companies with valuable data for research and development. Seeking external funding from international organizations, philanthropic organizations, and other sources could help overcome this challenge. External funding could help ensure that LMICs have access to the necessary resources required to implement and maintain AI systems in the healthcare sector.

Through the use of open-source software, AI technology costs can also be decreased. Open-source software can be accessed and modified by anyone, which can make it a more affordable option for healthcare facilities in LMICs. Together, healthcare workers and software developers can enhance the functionality and usability of AI systems, which can also foster cooperation and invention.

5.3 Challenge 2: Shortage of Professionals

The shortage of trained professionals capable of operating AI models is a challenge that is not unique to LMICs but is a global issue. Healthcare providers need specialized training to understand how to interpret and use the insights generated by AI systems effectively. The healthcare sector has a considerable need for skilled professionals who can effectively collaborate with AI, yet there is a scarcity of such qualified individuals at present (Hee Lee and Yoon 2021; Topol 2019).

There is also a dearth of experts with the skills required to run and manage AI models in healthcare in higher-income nations like the United States. According to a report by Burning Glass Technologies, there were over 7000 job postings for healthcare AI positions in 2019 in the United States alone (Burning Glass Technologies 2019). However, with growing demand, there is relatively limited number of available educational programs for training specialists in healthcare AI.

In LMICs, the shortage of trained professionals is even more pronounced. It may be challenging for healthcare workers in these environments to acquire the specialized skills required to run AI systems because they frequently have restricted access to educational tools and training opportunities. In addition, it is frequently difficult to find tools for ongoing professional growth and training, which can make it challenging to keep up with AI technology advancements. For example, in some LMICs, there may be only a few professionals trained in AI in healthcare in the entire country. This shortage of skilled professionals can limit the widespread adoption of AI technology in healthcare, even in facilities with the financial resources to acquire them.

5.4 Ways to Overcome Challenge 2: Shortage of Professionals

A potential approach to overcoming this challenge involves investing in training programs that help healthcare professionals in LMICs develop their proficiency in AI technology. This could be done in partnership with universities or research institutions that have expertise in AI and healthcare (Bajwa et al. 2021). These programs could include courses on data analytics, machine learning, and AI applications in healthcare.

An alternative solution involves adopting AI models that are easy to use and demand little training for operation. As an illustration, some businesses have created AI-powered software with a user-friendly interface that enables healthcare workers to use the tool with little to no training (Jacobs et al. 2021). These systems may be particularly beneficial in LMICs, where there is a shortage of skilled personnel to operate more complex AI models.

Furthermore, some initiatives are focused on developing AI tools that are specifically designed for low-resource settings, such as remote and rural healthcare facilities. The purpose of developing these tools is to provide medical professionals with decision support systems, empowering them to diagnose and treat patients effectively even in regions where medical resources are scarce or inadequate. These initiatives aim to make AI more accessible and user-friendly for healthcare professionals in LMICs.

Finally, international collaborations between LMICs and high-income countries can also be valuable for addressing the shortage of trained professionals in LMICs. Such collaborations could involve the exchange of knowledge and expertise, with high-income countries sharing their knowledge of AI and healthcare systems with LMIC. Such collaborations might also encompass partnerships among universities, research institutions, and hospitals from both high-income and LMICs to facilitate training and capacity-building efforts.

In conclusion, tackling the lack of skilled professionals who can operate AI models presents a crucial obstacle to implementing AI in healthcare within LMICs. This could pose a significant challenge for healthcare providers looking to fully utilize the benefits of AI technology.

5.5 Challenge 3: Infrastructure

To develop accurate and reliable AI models for healthcare, a large amount of medical data is required. This is because AI systems use statistical models to recognize patterns in the data they are trained on, and the larger the dataset, the more accurate the model can be. Wu et al. proposed an artificial intelligence (AI) system based on deep convolutional neural networks (DCNN) to classify breast cancer screening exams using mammography. The system was trained and evaluated on over 200,000 exams, which incorporated over 1,000,000 images. This demonstrates an example of AI requiring big data since the large dataset was necessary to train and test the system’s accuracy in classifying breast cancer exams accurately. The performance of their network achieved an AUC of 0.895 in predicting whether there is a cancer in the breast, when tested on the screening population and the result was compared to 14 radiologists reading results. The study found that the hybrid model, averaging the probability of malignancy predicted by a radiologist with a prediction of their neural network, is more accurate than either of the two separately. For λ = 0.510, hybrids between each reader and the model achieved an average AUC of 0.891 (Wu et al. 2020).

The quality of the data, in addition to its amount, is the other key factor of enhancing accuracy of AI algorithms. Incomplete or biased data can lead to inaccurate or misleading results. To gather, store, and handle high-quality medical data for AI applications, it is crucial to have a solid data infrastructure in place. When it comes to obtaining and storing medical data, LMICs may not have enough resources to meet the demand for significant quantities of high-quality data. For instance, it may be challenging to obtain and analyze the data in some areas because usually medical records are on paper. Additionally, there may be cultural or privacy concerns that limit the sharing of medical data, further limiting the amount of data available for AI applications (Bak et al. 2022).

However, there can be issues with data access and accuracy even in high-income nations with well-established healthcare systems. Medical data, for instance, may be kept in various forms by various healthcare systems, making it challenging to aggregate and evaluate. Furthermore, there might be issues with data security and patient privacy that restrict the exchange of medical information between various organizations. Efforts to improve data infrastructure and promote the responsible sharing of medical data will be essential aspects to unlocking the full potential of AI in healthcare.

5.6 Ways to Overcome Challenge 3: Infrastructure

Strengthening infrastructure remains a crucial step towards addressing challenges related to data availability and quality in healthcare AI (Obermeyer and Emanuel 2016). A dependable and well-structured system for collecting, storing, and transmitting medical data is vital to guarantee that substantial volumes of high-quality data are accessible for training AI models (Murdoch and Detsky 2013).

One strategy for achieving this is through the creation of cloud-based systems accessible to healthcare providers from various locations. This approach permits medical data to be securely stored and accessed from a centralized location, simplifying aggregation and analysis. Furthermore, cloud-based systems offer scalability, accommodating increasing amounts of medical data over time (Chassagnon et al. 2020).

High-speed internet connectivity represents another essential aspect of infrastructure investment, enabling the real-time transmission of medical data (Schwamm et al. 2020). This allows healthcare providers to rapidly and effortlessly view and exchange medical data online, regardless of their location. In remote or underdeveloped regions, where medical resources might be limited (Wosik et al. 2020). By enhancing data infrastructure, we can ensure the availability of high-quality medical data for AI applications, resulting in more precise and trustworthy AI models.

5.7 Challenge 4: Ethical Concerns

The deployment of AI in healthcare, particularly in LMICs, presents numerous ethical concerns that require attention. One of the primary issues is privacy, as sensitive patient information needs protection from unauthorized access and usage. Patient data is generally considered confidential, and its disclosure or misuse can result in serious repercussions for individuals, such as discrimination, stigma, and even health risks. In high-income countries, strict regulations like HIPAA in the United States and GDPR in Europe safeguard patient privacy. However, in LMICs, there may be insufficient regulatory frameworks to protect patient data, posing a significant risk to patient privacy (Price and Cohen 2019).

Another ethical concern is bias, as AI models may learn from biased data, leading to discriminatory results. This problem is especially relevant in LMICs, where diverse data for training AI models may be scarce. In high-income countries, efforts to reduce bias in AI models include using diverse data and testing for biases during model development. However, in LMICs, addressing this issue effectively may be hindered by limited resources and expertise (Rajkomar et al. 2018). Accountability is another ethical consideration when employing AI in healthcare. If AI models make incorrect decisions, mechanisms should be in place to hold responsible parties accountable. This is particularly pertinent in LMICs, where AI usage in healthcare may be relatively new, and regulatory frameworks for accountability may be lacking.

Developing ethical guidelines and regulatory frameworks for AI in healthcare is especially crucial in LMICs, where resources for addressing these issues might be limited. In high-income countries, efforts to establish ethical guidelines and regulatory frameworks for AI in healthcare are already underway (Fiske et al. 2019). For example, the European Union’s General Data Protection Regulation (GDPR) includes rules for the security of personal data, encompassing medical data. In the United States, regulatory organizations like the Food and Drug Administration (FDA) oversee the use of AI in healthcare.

While effectively addressing these issues in LMICs may be challenging due to limited regulatory frameworks and resources, establishing robust regulatory frameworks and ethical guidelines can help ensure that AI is employed ethically and responsibly to benefit patients and healthcare systems. Careful planning and development of suitable frameworks are vital for responsible and effective AI utilization in healthcare.

5.8 Challenge 5: Accountability and Shifts in Responsibility

The increasing role of AI systems in healthcare raises concerns about how the shift in responsibility affects various stakeholders, including medical professionals, AI developers, and patients. The growing reliance on AI could lead to clinicians becoming overly dependent on these systems, potentially causing a decline in their own skills or personal connections with patients (Shortliffe and Sepúlveda 2018). This shift might also create new obligations for AI developers, who could gain significant influence on healthcare and should be responsible for creating safe, useful AI systems while responsibly shaping public views on health (Price and Cohen 2019).

Moreover, the expansion of AI in healthcare raises concerns about accountability, as it is currently unclear who should be held responsible if a thoroughly clinically validated model makes mistakes (Vayena et al. 2018). Traditionally, doctors are held liable when they deviate from the standard of care, and patient injury occurs. However, as the standard of care evolves to incorporate AI tools, there will be a strong medicolegal incentive for doctors to follow AI recommendations, which can create a dilemma when AI recommendations conflict with standard practice.

5.9 Ways to Overcome Challenge 5: Accountability and Shifts in Responsibility

To overcome this challenge, a human-centered approach is essential. One way to address this issue is to offer patients the option to choose between an AI-based tool or a human practitioner. This allows patients to have a say in their own healthcare while promoting transparency and accountability in the use of AI. By respecting patients’ preferences and values, healthcare providers can build trust and improve patient outcomes (Gerke et al. 2020).

Lastly, as people’s beliefs may still pose obstacles in trusting AI, it is essential to approach the use of AI systems in healthcare as a continuous learning process. As new challenges and ethical concerns arise, remaining open to feedback and collaboration is crucial to finding effective solutions (Tachkov et al. 2022). By fostering a culture of collaboration and continuous education, healthcare providers, AI developers, and patients can work together to optimize the use of AI systems in healthcare and ensure that accountability and responsibility are shared among all stakeholders.

5.10 Other Challenges and Ways to Overcome These Barriers

Addressing the challenges posed by AI in healthcare for LMICs is vital, as the potential benefits of AI are immense. One of the primary obstacles is the language barrier, which can limit the effectiveness of AI algorithms due to a lack of local language data for training purposes. Cultural beliefs and traditions can also affect patients’ willingness to use AI tools. In rural areas, patients may be hesitant to utilize AI tools due to unfamiliarity with technology and a preference for face-to-face interactions with healthcare providers. Overcoming this reluctance is essential to fully realize the benefits of AI in healthcare. To address these issues, it is crucial to create culturally and linguistically relevant AI technologies that cater to the specific needs of LMICs (Meskó et al. 2017). Collaborating with local communities, healthcare providers, and research institutions can aid in developing AI tools that align with cultural beliefs and values. Additionally, establishing large open-source databases in local languages for training AI algorithms can improve the accuracy and applicability of AI models in these regions (Liu et al. 2019). Healthcare providers and institutions can contribute data, while technology companies can provide the necessary expertise and tools to build and maintain these databases. Developing multilingual AI models or models that can adapt to new languages with minimal training data can help ensure that AI models are effective in diverse linguistic contexts (Yang et al. 2022).

The successful implementation of AI in LMICs requires a multi-stakeholder approach that emphasizes collaboration, innovation, and equity. By working together, we can harness the power of AI in healthcare and improve health outcomes for all.

6 Mitigating Risks and Addressing Ethical Concerns in AI-Driven Healthcare

As the future of all-encompassing AI in healthcare offers tremendous potential, we must exercise caution and vigilance to ensure the responsible and ethical harnessing of this transformative capability. As we strive for this promising future, there are several critical considerations to bear in mind to mitigate potential hazards and prevent unintended consequences.

First, the privacy and security of patient data must be a primary concern as we incorporate AI into healthcare systems. Since AI algorithms depend on extensive data, we must establish strong data protection measures to prevent unauthorized access, misuse, or breaches of confidential patient information. This will call for ongoing vigilance and the creation of advanced cybersecurity protocols, particularly in LMICs where existing data infrastructure might be less secure.

Second, we must address the potential for bias in AI algorithms, as they are only as impartial as the data on which they are trained. Ensuring that AI models are trained on diverse and representative datasets is crucial to prevent discriminatory outcomes and the perpetuation of health disparities. This will require continuous monitoring and evaluation of AI-driven healthcare solutions to identify and correct any biases that may emerge.

Furthermore, as AI becomes increasingly integrated into healthcare decision-making, we must not lose sight of the importance of human interaction in medicine. Healthcare professionals must continue to play a central role in patient care, using AI as a tool to augment their expertise rather than supplanting the essential human connection that underpins compassionate care.

Lastly, we must foster global collaboration and the equitable distribution of AI-driven healthcare solutions to ensure that the benefits of this technology are available to everyone, regardless of their location or socioeconomic status. This will necessitate a focused effort to invest in capacity-building, infrastructure development, and local expertise in LMICs, creating an inclusive global healthcare community that embraces the transformative power of AI while addressing its inherent challenges.

By remaining aware of these considerations and working together to tackle potential risks, we can responsibly harness the power of all-encompassing AI in healthcare and create a future where the bright possibilities are realized, and the potential pitfalls are expertly navigated.

7 Future Directions for Ubiquitous AI in Healthcare in LMICs

7.1 Expanding AI in Primary Care Settings

The potential for AI in primary care settings in LMICs offers a valuable opportunity to enhance healthcare access and outcomes in remote and underserved regions. Primary care practitioners are vital in preventive care, early diagnosis, and managing chronic conditions, making them ideal candidates for AI integration.

AI-driven diagnostic tools empower primary care practitioners to deliver accurate and efficient diagnoses, even in regions where healthcare professionals are scarce. For instance, AI-powered mobile apps that use machine learning algorithms to analyze images or symptoms could help diagnose conditions like diabetic retinopathy or skin cancer.

Telemedicine integration can enable primary care practitioners to provide remote consultations and follow-up care to patients living far from healthcare facilities. This technology also allows for collaboration with specialists and other healthcare professionals, improving patient care quality.

Electronic health records (EHRs) play an essential role in AI integration. EHRs store patient data and facilitate AI-powered algorithms to create personalized treatment plans, improving treatment outcomes, reducing adverse event risks, and allowing for more efficient healthcare resource utilization.

7.2 Enhancing Public Health Surveillance and Response

Enhancing public health surveillance and response is another promising future direction for AI in healthcare in LMICs. AI-powered tools can be used to monitor disease patterns, forecast outbreaks, and optimize resource allocation, enabling a more efficient and targeted response to public health emergencies.

For example, during the COVID-19 pandemic, AI algorithms have been used to forecast the spread of the virus, enabling public health officials to allocate resources and plan interventions accordingly. AI-powered surveillance systems can also monitor social media and other online platforms to detect outbreaks in real-time, enabling a more rapid response to public health emergencies.

In addition, AI can be used to develop disease surveillance systems that enable early detection and response to disease outbreaks. AI-powered tools can be used to analyze large amounts of data from various sources, including EHRs, disease registries, and public health databases, to identify patterns and predict the likelihood of outbreaks.

Furthermore, AI can be used to optimize resource allocation during public health emergencies. For example, AI-powered algorithms can be used to predict which healthcare facilities are likely to experience a surge in demand, enabling public health officials to allocate resources more efficiently.

Envisioning the integration of AI-powered public health surveillance and response tools in the future of healthcare in LMICs presents an opportunity to improve the outcomes.

7.3 Strengthening Healthcare Workforce Capacity

AI-driven solutions can enhance healthcare professionals’ skills, minimize knowledge disparities, boost diagnostic precision, and ultimately lead to better patient outcomes. AI-facilitated training platforms offer healthcare professionals access to immersive virtual training environments, allowing them to acquire hands-on experience without costly, time-consuming in-person training. AI can further contribute to the development of decision support systems that guide healthcare professionals in making precise diagnoses and treatment choices, thus decreasing error risks, and enhancing patient outcomes (Spatharou et al. 2020).

Furthermore, AI-driven tools can supply healthcare professionals with the latest medical information, ensuring they remain informed about recent medical research and best practices (Davenport and Kalakota 2019). AI solutions can also automate administrative tasks, freeing up healthcare professionals to focus on patient care rather than paperwork.

7.4 Collaborative Research and Development

Collaborative research and development play a crucial role in the future of AI in healthcare for LMICs. Considering the unique challenges and contexts of LMICs, fostering international collaboration among governments, academic institutions, and industry partners is essential for creating AI-based healthcare solutions tailored to LMICs’ specific needs.

By joining forces in research and development, stakeholders can capitalize on each other’s strengths to create more effective and efficient AI healthcare solutions. Governments can offer funding and establish regulatory frameworks to back AI research and development, while academic institutions can contribute their expertise in AI algorithms and data analysis. Meanwhile, industry partners can supply resources and know-how in developing and commercializing AI-driven healthcare solutions.

Additionally, international collaborations can address LMICs’ challenges related to data availability and quality. Cross-border data sharing enables more comprehensive and accurate health trend analyses, which, in turn, facilitates the development of more efficient AI-driven healthcare solutions (Wirtz et al. 2017; Vayena et al. 2018).

7.5 Long-Term Sustainability and Scalability

Although AI has the potential to revolutionize healthcare in LMICs, a comprehensive approach addressing these settings’ unique challenges and opportunities is necessary to ensure long-lasting success.

Sustainable funding remains a key consideration. AI-based healthcare solutions demand significant investments in research, development, maintenance, and support. It is vital to establish funding models that prioritize LMICs’ needs, providing continuous, long-term backing for AI-driven healthcare initiatives. Infrastructure development is another crucial factor. To enable widespread AI adoption in LMICs’ healthcare systems, investments in infrastructure, such as internet connectivity, computing resources, and data storage, are necessary. Additionally, creating user-friendly interfaces and mobile applications that are easily accessible by healthcare professionals in remote or underserved areas is essential.

Furthermore, continuous education and training play a critical role in the long-term success and scalability of AI-driven healthcare solutions in LMICs. Healthcare professionals require ongoing training and education to effectively incorporate AI technologies into their workflows and fully utilize these solutions’ potential. Also, addressing language and cultural barriers through continuous education and training ensures that AI-driven healthcare solutions remain accessible and effective across diverse settings (Topol 2019; Carvalho et al. 2022).

8 Embracing the Future: A Vision of Ubiquitous AI for Universal Healthcare

As we gaze into the future, we imagine a world where AI becomes an integral part of healthcare delivery, transcending geographical boundaries and socioeconomic barriers to create a reality where healthcare is genuinely universal. In this envisioned future, AI-driven innovations will narrow the gap between LMICs and high-income countries, empowering LMICs to overcome traditional developmental challenges and achieve equal access to quality healthcare for everyone.

In this inspiring future, AI will act as a catalyst for proactive, preventive healthcare, benefiting individuals in both LMICs and high-income countries through real-time health monitoring, personalized lifestyle advice, and early interventions. This fundamental change will enable healthcare systems to shift from reactive care to predictive and preventive care, ultimately reducing the burden of disease and enhancing overall well-being.

We imagine healthcare providers worldwide connected via AI-powered telemedicine platforms, working together to deliver expert care and share knowledge across borders. This interconnected healthcare ecosystem will encourage innovation, fuel medical breakthroughs, and enable the rapid spread of best practices, ensuring that state-of-the-art medical advances are accessible to everyone, regardless of their location. In this future, AI will be an indispensable tool in addressing worldwide health crises, such as pandemics and emerging infectious diseases. AI-driven early warning systems will enable prompt identification of disease outbreaks, while AI-powered epidemiological models will guide evidence-based interventions, optimizing resource allocation and enabling a coordinated global response.

Moreover, AI will become a driving force for social change, dismantling barriers to healthcare access and diminishing health disparities in LMICs. AI-driven healthcare solutions will be culturally and linguistically inclusive, respecting local customs and beliefs while adapting to unique contexts to ensure fair and effective care for diverse populations. This future is not merely aspirational but achievable. By harnessing the power of AI, fostering international collaboration, and investing in the development of human-centered, ethical AI solutions, we can create a world where healthcare is no longer a privilege but a fundamental right for all. Embracing pervasive AI in healthcare will lay the groundwork for a brighter, healthier future, where no one is left behind in the pursuit of well-being and an improved quality of life.

9 Conclusion: Harnessing the Potential of Ubiquitous AI to Transform LMIC Healthcare

In conclusion, this analysis has highlighted the considerable potential of ubiquitous AI in revolutionizing healthcare for LMICs. We have explored how AI integration can dramatically enhance disease diagnosis, personalized treatment, drug development, and overall healthcare delivery, even in resource-limited settings. We have also showcased the transformative impact of AI-driven precision medicine, wearable devices, electronic health records, and imaging techniques that can bring meaningful benefits to LMICs. Success stories from Rwanda, Armenia, Pakistan, and India serve as inspiring examples for other LMICs to follow.

Despite these promising prospects, we have acknowledged the challenges and constraints of implementing AI in healthcare within LMICs, including cost barriers, the scarcity of skilled professionals, and data quality and accessibility issues. To overcome these hurdles, we have proposed solutions such as public-private partnerships, external funding, open-source software, international collaborations, and investments in infrastructure and training.

As we look to the future, we encourage researchers and practitioners to pursue AI applications in primary care, diagnostics, telemedicine, and public health surveillance. These innovative technologies have the potential to extend healthcare access and improve outcomes in remote and underserved areas, optimize resource distribution during public health emergencies, and facilitate more efficient and targeted responses to disease outbreaks.

In conclusion, it is crucial for all stakeholders—governments, academic institutions, and industry partners—to collaborate and invest in the creation and implementation of AI-driven healthcare solutions tailored to the unique needs of LMICs. By focusing on sustainable funding models, infrastructure development, user-friendly interfaces, and continuous education and training, we can unlock the full potential of AI in healthcare.

Together, we can pave the way for a brighter future where the power of AI transforms the lives of millions of individuals in low- and middle-income countries, fostering hope, health, and prosperity for generations to come.