Profile Pic Paul Minnis

The use of AI in biomarkers, CT omics and pathology interpretation

Monday, April 22, 2024

Dr Paul Minnis is a specialist in Interstitial Lung Disease with interests in the inflammatory mechanisms driving fibrotic lung disease and the application of technology in the diagnosis and management of chronic disease. His MD focused on the assessment of novel sensors investigating the potential for connected health solutions in chronic disease at the Insight Centre for Data Analytics UCD.

 

The terms Artificial Intelligence (AI) is an umbrella concept for a collection of technologies including machine learning involving neural networks and deep learning, natural language processing, rule-based expert systems and robotic process automation. It can be defined as the overarching theory, development, and simulation of human intelligence by computer systems covering tasks such as perception, reasoning, decision making, language processing, and the display of knowledge or information.

Research into AI in medicine has accelerated markedly over the last decade. This has primarily been driven by technological advancements in computational power and the digital revolution in medicine and healthcare information resulting in the increased acquisition and availability of large volumes of different types of intertwined data including genomics, biomarkers, medical imaging, and electronic health records.

The field of respiratory medicine is well represented, AI promises to support clinical decision-making processes by optimising diagnosis and treatment strategies of heterogeneous and complex chronic respiratory diseases. Its capabilities extend to predicting exacerbation risk, disease progression and mortality, providing healthcare professionals with valuable insights for more effective care. Further advantages include the facilitation of AI evaluation in isolated areas or underserved locations which could potentially transform patient outcomes by providing more timely and accurate diagnosis.

The main applications of AI thus far have been the interpretation of thoracic imaging, lung pathology, and physiological data such as pulmonary function tests as well as the evaluation of potential biomarkers.

Within respiratory medicine, the main applications of AI thus far have been the interpretation of thoracic imaging, lung pathology, and physiological data such as pulmonary function tests as well as the evaluation of potential biomarkers. There are already several research studies suggesting that AI can perform as well as or better than humans at key healthcare tasks, such as diagnosing disease.

How is AI currently being considered/used in the prevention/treatment/management of ILD?

Diagnosis and treatment of disease has been a focus of AI since its inception in the 1970s. It is playing an increasingly significant role in various aspects of healthcare management, including the prevention, treatment, and management of Interstitial Lung Disease (ILD).[i] AI can aid early detection and diagnosis, allow for image analysis and quantification, perform predictive analytics, accelerate drug discovery and development, facilitate personalized treatment planning as well as remote monitoring.

AI algorithms can analyse medical imaging data such as chest X-rays and CT scans to detect patterns indicative of ILD at an early stage. These algorithms can often identify subtle abnormalities that might be overlooked by human observers, leading to earlier diagnosis and intervention. AI can assist radiologists in quantifying the extent and severity of lung involvement in ILD by automatically measuring parameters like fibrosis extent, ground-glass opacities, and honeycombing on imaging studies.[ii] This helps in disease staging and monitoring progression over time. Machine learning models can analyse patient data, including clinical records, imaging studies, and biomarkers, to predict disease progression and prognosis in individuals with ILD. This enables healthcare providers to tailor treatment plans and interventions based on individual patient risk profiles.

Overall, AI holds great promise in revolutionising the early detection, facilitating diagnosis, and management of ILD by improving accuracy, efficiency, and personalised care delivery.[iii] However, ongoing research and collaboration between healthcare providers, researchers, and technology developers are essential to realise the full potential of AI in this field.

How is AI being used to analyse biomedical markers?

AI plays a crucial role in analysing biomedical markers by extracting actionable insights from complex data, facilitating biomarker discovery, diagnostic modelling, prognostication, and personalized medicine across various healthcare domains.

AI algorithms can integrate and analyse large-scale biomedical datasets, including genomics, proteomics, metabolomics, and clinical data, to identify patterns, correlations, and associations that might not be apparent through traditional statistical methods.

AI algorithms can integrate and analyse large-scale biomedical datasets, including genomics, proteomics, metabolomics, and clinical data, to identify patterns, correlations, and associations that might not be apparent through traditional statistical methods. This helps in understanding the complex relationships between different biomarkers and disease states. Machine learning techniques, such as feature selection and classification algorithms, are used to identify novel biomarkers associated with specific diseases or conditions. AI can analyse multi-omic data to uncover biomarkers indicative of disease risk, progression, or treatment response, facilitating early diagnosis and personalised medicine.

What are some of the benefits/risks of using AI in this way?

Using AI for the analysis of biomedical markers offers numerous benefits, but it also comes with certain risks. Benefits include improved accuracy, early detection and diagnosis, development of personalised treatment plans based on individual biomarker profiles, and data-driven insights. AI has the potential to uncover hidden patterns and associations within biomedical data, leading to new discoveries, hypotheses generation, and advancements in biomedical research.

Possible shortfalls include poor quality of training data such as underrepresentation of certain demographic groups or overrepresentation of specific populations, leading to biased or inaccurate predictions and recommendations. Algorithmic bias may reflect and potentially exacerbate existing disparities in healthcare outcomes. AI models, particularly deep learning models, are often regarded as "black boxes" due to their complex nature, making it challenging to interpret how they arrive at specific predictions. Lack of transparency can undermine trust and raise ethical concerns, especially in critical healthcare decisions. Excessive reliance on AI-driven analyses may lead to complacency among healthcare professionals, diminishing their clinical judgment and critical thinking skills. It's crucial to view AI as a tool to augment rather than replace human expertise and judgment.

Addressing these risks requires careful consideration of ethical, legal, and social implications, along with ongoing efforts to improve data quality, transparency, accountability, and equity in AI-driven healthcare applications.[iv]

What value does using AI in the analysis of omics data provide?

Omics data, such as genomics, transcriptomics, proteomics, and metabolomics, generate massive amounts of complex data that traditional analytical methods struggle to handle. AI techniques excel at processing and analysing such high-dimensional data, extracting meaningful patterns, and uncovering hidden relationships among different molecular components.

By integrating multiple omics layers and clinical data, AI algorithms can pinpoint molecular signatures indicative of specific biological states or pathological conditions, facilitating early diagnosis and personalised medicine approaches. AI algorithms provide deeper insights into the underlying biological mechanisms driving disease pathogenesis and progression by identifying key molecular pathways, regulatory networks, and genetic variations associated with specific phenotypes or clinical traits. By elucidating the molecular basis of diseases, AI-driven omics analysis accelerates biomedical research and drug discovery efforts.

AI analysis of omics data enhances our understanding of biological systems, accelerates biomedical research, and transforms healthcare delivery by enabling precision medicine approaches tailored to individual patients' molecular profiles. By leveraging AI's analytical power and predictive capabilities, researchers and clinicians can unlock new insights into disease mechanisms, improve diagnostic accuracy[v], and optimise treatment outcomes in diverse clinical settings.

Are there any patient privacy concerns in using AI for pathology interpretation?

The use of AI in biomedical research and healthcare raises concerns about data security, confidentiality, and the risk of unauthorised access or data breaches.

Yes, there are patient privacy concerns associated with using AI for pathology interpretation, primarily related to the handling and analysis of sensitive medical data. Biomedical data are highly sensitive and subject to strict privacy regulations. The use of AI in biomedical research and healthcare raises concerns about data security, confidentiality, and the risk of unauthorised access or data breaches. The regulatory landscape for AI in healthcare is still evolving, posing challenges related to validation, standardisation, and compliance with regulatory requirements. Finally, the future of AI-related medical applications are reliant on ensuring the safety, efficacy, and ethical use of AI technologies in healthcare and this requires robust regulatory frameworks and oversight.

 

References

[i] Huang Y, Ma SF, Oldham JM, Adegunsoye A, Zhu D, Murray S, Kim JS, Bonham C, Strickland E, Linderholm AL, Lee CT. Machine Learning of Plasma Proteomics Classifies Diagnosis of Interstitial Lung Disease. American Journal of Respiratory and Critical Care Medicine. 2024 Feb 29(ja).

[ii] Naik N, Hameed BM, Shetty DK, Swain D, Shah M, Paul R, Aggarwal K, Ibrahim S, Patil V, Smriti K, Shetty S. Legal and ethical consideration in artificial intelligence in healthcare: who takes responsibility?. Frontiers in surgery. 2022 Mar 14;9:266.

[iii] Handa T. The potential role of artificial intelligence in the clinical practice of interstitial lung disease. Respiratory Investigation. 2023 Nov 1;61(6):702-10.

[iv] Walsh SL, De Backer J, Prosch H, Langs G, Calandriello L, Cottin V, Brown KK, Inoue Y, Tzilas V, Estes E. Towards the adoption of quantitative computed tomography in the management of interstitial lung disease. European Respiratory Review. 2024 Jan 31;33(171).

[v] Dack E, Christe A, Fontanellaz M, Brigato L, Heverhagen JT, Peters AA, Huber AT, Hoppe H, Mougiakakou S, Ebner L. Artificial intelligence and interstitial lung disease: Diagnosis and prognosis. Investigative radiology. 2023 Aug 1;58(8):602-9.