AI in Healthcare Quality: Advances and Ethical Concerns

Disease Diagnosis

Medical imaging techniques such as X-rays, Magnetic Resonance Imaging (MRI), and Doppler scans generate image data that is crucial for diagnosing diseases. This imaging data helps healthcare professionals identify and assess various conditions with greater accuracy [12]. Despite the availability of advanced imaging data and comprehensive test reports, achieving precise disease diagnosis remains challenging without extensive expertise and up-to-date research knowledge. The complexity of medical conditions and the need for expert interpretation underscore the limitations of current technology alone [9]. AI bridges these gaps through transfer learning, where neural networks are trained using extensive image datasets curated by radiologists. Although AI is built upon existing knowledge and human intelligence, it utilizes this foundational data to update and integrate new information continuously. This process enables AI systems to adapt and evolve, enhancing their ability to provide insights and recognize patterns, which ultimately leads to improved accuracy and reliability in disease diagnosis. This complements medical professionals’ expertise and helps overcome some limitations of traditional diagnostic approaches [12]. For example, AI system trained on over 130,000 images of skin lesions has learned to identify key features linked to skin cancer, achieving diagnostic accuracy comparable to that of professional dermatologists [13].

DeepMind (Google) is actively researching AI for disease diagnosis, focusing on developing advanced algorithms to improve diagnostic accuracy and efficiency [9].

Similarly, IBM Watson researchers are evaluating the effectiveness of AI in diagnosing breast cancer, aiming to enhance early detection and treatment outcomes. Both efforts represent significant strides in integrating AI into medical practice, with the potential to transform how diseases are diagnosed and managed through advanced data analysis and pattern recognition [14]. Peter et al. surveyed the diagnosis and grading of malignant and benign prostate cancers, contributing valuable insights into the use of AI for cancer diagnosis [15]. Hasan et al. achieved a remarkable 99.68% accuracy in classifying COVID-19, pneumonia, and healthy individuals using advanced AI techniques [16]. AI facilitates radiologists primarily in the following aspects [17].

• Accurate diagnoses with operational advantages, including reduced time consumption and increased consistency.
• Intuitive and user-friendly application, ensuring easy data access even after integrating AI systems, such as the incorporation of BoneXpert into Picture Archiving and Communication Systems (PACS).
• Decrease in healthcare expenses.

Precision Medicine

An individual’s quality of life in terms of health depends not only on early diagnosis and detection but also on effective treatment. Historically, healthcare practitioners often provided generalized treatments for common symptoms, sometimes overlooking individual variations and side effects [18]. Advancements in technology now enable the provision of evidence-based treatments and the exploration of alternative solutions for patients experiencing adverse effects. Precision medicine analyses diagnostic data, genomic profiles, lifestyle, and epigenetics to predict individual medication responses. This approach enables the development of personalized treatments, tailored to each patient by considering all relevant factors, ultimately enhancing both health outcomes and treatment efficiency [19, 20]. In healthcare, generative AI is proving its ability to go beyond traditional knowledge constraints and promote innovation. One of its key contributions is in drug discovery, where it simulates molecular interactions and predicts new compounds, thereby accelerating the development of novel therapies. Researchers are actively exploring the development of digital twins that simulate individual components or entire systems of a person, such as organelles, organ systems, or receptors, to enable personalized medical care [21]. These have been evident in the development of a digital twin of the liver to study its functions, disease states, and drug responses [22]; guardian angels which are customized to mirror the neology of European patients [23] and also in orthodontic treatments [24]. While the essence of Generative intelligence (GI) lies within human cognition, the potential for AI systems to assist in innovative processes in healthcare is a major feat.

Telemedicine

Enhancing healthcare quality primarily focuses on improving patient well-being. Frequent hospital visits for health monitoring can be challenging for patients with limited mobility. Telemedicine addresses this issue by enabling remote management of patient conditions through real-time interventions using point-of-care devices, wearables, and sensors, thus reducing the need for frequent hospital visits. Advances in information technology and global connectivity have facilitated the growth of telemedicine, making it a viable solution for continuous patient care [25]. During the COVID-19 pandemic, telemedicine played a crucial role by facilitating communication between patients and doctors, raising public awareness, and minimizing disease transmission risks. Additionally, AI robots contributed to reducing infection risks, especially during pandemics and epidemics, by supporting two-way communication between physicians and patients and serving as personal trainers to help maintain physical fitness [26]. Amidst the COVID-19 pandemic, CPAIS (COVID-19 Pandemic AI system) was developed to forecast the trends, and variations of COVID-19 globally (databases from approx.171 countries). The datasets of performances of the management of pandemic responses from countries like United States, India, France, Taiwan etc., were utilised to track the infection rates, and healthcare demands that leverages the predictive capabilities and provide information of health strategies globally, allocating resources in various countries [27]. Voice assistants assist the elderly and visually impaired by enhancing mental health, reducing depression and stress, and serving as companions [28].

Bias, Safety and Ethical Consideration

While integrating AI systems in primary healthcare offers numerous benefits, it also presents challenges related to bias and ethics. Although these systems can access vast amounts of data, there is often a lag in continuously updating this data, resulting in outdated clinical practices—especially in rapidly evolving research. Consequently, healthcare providers may struggle to keep AI models current due to inadequate tools, leading to potential gaps in care. Additionally, the lack of technical expertise can hinder the effectiveness of AI in patient management [29]. Frequent software updates can be time-consuming and may complicate compatibility and user- friendliness [30]. To address these issues, adopting cloud- based infrastructure, automated systems, and collaborative networks with data scientists can help establish frameworks for regular updates and create user-friendly applications for healthcare practitioners.

The datasets used to train AI models often originate from specific populations, which can introduce bias into their applicability. Consequently, these AI technologies tend to be utilized mainly by elite practitioners, while underrepresented populations are frequently overlooked. This highlights the importance of ensuring equitable healthcare by expanding the diversity of data collected from various populations [31]. Medical ethics have been established since the 1950s, emerging as a response to the aftermath of World War II, with a focus on prioritizing the well-being of individuals [32]. The principles of medical ethics encompass obtaining patient consent, ensuring safety through physician care, protecting patient privacy, and respecting the patient’s morals [33]. The advancement of AI in healthcare presents numerous benefits, but it also necessitates careful consideration to ensure patient safety. Patients’ healthcare records contain confidential information, including genetic profiles and medical histories [34, 35]. The IEEE Global Initiative stresses that ethical concerns in AI and autonomous systems can be easily overlooked or oversimplified in technical fields. It calls for comprehensive and effective ethics education within engineering programs, integrated across coursework and practical training. The objective is to ensure that ethics and human rights are seamlessly included in the design process of AI technologies [36]. Consequently, it is essential to develop and program AI assistants and robotics within established ethical frameworks to ensure their responsible use [33].