How Augmented AI Will Power Automated Healthcare
Authors: Zain Khalpey, MD, PhD, FACS, Jessa Deckwa, BS, and Amina Khalpey PhD
Introduction
Augmented AI is already being used in various applications within the healthcare industry, and its potential for future growth is vast. Advances in AI software and hardware, especially deep learning algorithms and the graphics processing units (GPUs) that power their training, have led to a recent and rapidly increasing interest in medical AI applications (1). Here are some specific examples of how augmented AI is being used in healthcare:
1. Medical Imaging
One of the most widely used applications of augmented AI is in medical imaging, which is essential for accurate diagnosis and treatment planning. AI systems can analyze large amounts of data from medical images, such as X-rays, CT scans, and MRIs, and identify patterns and anomalies that may be difficult for human doctors to spot.
Advancements in Medical Image Reading with AI
Artificial intelligence algorithms were developed to assess their performance in the clinical setting of mammograms for breast cancer screening, by first curating a large representative dataset from the UK and a large, enriched dataset from the USA. Despite early screening to detect an early stage of breast cancer development, these screens are affected by high false positive and false negative rates (2). One algorithm showed an absolute reduction of 5.7% and 1.2% (USA and UK) in false positives and 9.4% and 2.7% in false negatives (3).
In a retrospective manner the algorithm was proven effective but in order to understand the practicality it was compared to actual radiologist readings. In an independent study of six radiologists, the AI system outperformed all of the human readers: the area under the receiver operating characteristic curve (AUC-ROC) for the AI system was greater than the AUC-ROC for the average radiologist by an absolute margin of 11.5% (3). A simulation in which the AI system participated in the double-reading process that is used in the UK and found that the AI system maintained non-inferior performance and reduced the workload of the second reader by 88% (2).
Disease Classification and AI
In patients affected with lung cancer, there is much debate over low-dose CT (LDCT) scans for early detection and risk benefit analysis to determine the best approach for early screening methods. After reviewing 96,559 subjects, LDCT screening significantly reduced lung cancer mortality, though not overall mortality, with women appearing to benefit more than men (4). If we applied a similar method to lung cancer as shown in breast cancer, we could truly understand the needed parameters for early detection.
The study showed the pooled false positive rate was 8% (95% CI, 4-18); subjects with false positive results had < 1 in 1000 risk of major complications following invasive diagnostic procedures (4). Not only could we reduce the false positive rate with AI but could also impact the false negative rate as seen in the breast cancer study. The most valid estimates for overdiagnosis and significant incidental findings were 8.9% and 7.5%, respectively (4). Another study reviewed Chest CT images have been able to distinguish lung adenocarcinoma, squamous cell carcinoma, small cell lung cancer, and other types based on the CNN algorithm model to classify lung cancer pathological images, which showed good application prospects (5). By detecting diseases at an early stage, AI can help improve patient outcomes and reduce healthcare costs.
2. Personalized Healthcare
Augmented AI can be used to develop personalized treatment plans based on individual patient data. For example, by analyzing a patient’s genetic makeup, medical history, and lifestyle factors, AI can identify treatments that are most likely to be effective for that particular patient.
Genetic Make-up and AI
In genomic data, a study was performed using unsupervised machine learning to determine different subclasses of acute myeloid leukemia (AML) as opposed to the historical subdivision of primary/de novo AML and secondary AML has shown to variably correlate with genetic patterns (6). Although the heterogeneity inherent in the genomic changes across nearly 7000 AML patients was too vast for traditional prediction methods, machine learning methods allowed for the definition of novel genomic AML subclasses, indicating that traditional pathomorphological definitions may be less reflective of overlapping pathogenesis. This approach is already being used in oncology, where AI is being used to analyze patient data to identify targeted therapies for cancer treatment.
Past Medical History and AI
Another study observed a different approach to predicting diabetes using lifestyle-related indicators as opposed to BMI, unhealthy diet, alcohol, smoking, and physical inactivity. Using the NHANES data of 17,833 respondents and 18 lifestyle variables for model training a study compared CATBoost, RF, LR, and XGBoost models in predicting diabetes (7). Among these, the CATBoost model achieved an accuracy of 82.1% and an AUC of 0.83 (7). These results indicated the proposed model can better predict diabetes using lifestyle-related indicators.
As AI becomes more accessible the applications in the medical field will continue to progress. In order to avoid bias in our models, we must continue using aggregated data and increasing the data parameters to ensure better inclusion and normalization.
3. Administrative Tasks
Needs for AI Integration in General Practice
Augmented AI can also be used to improve the efficiency of administrative tasks, which can free up healthcare professionals to focus on patient care. General practitioners (GPs) must diagnose, monitor, and manage treatment plans, as well as provide preventative medicine and screening – frequently under pressing time constraints due to the need to visit other patients or meet laboratory demands (8). A significant amount of a GP’s time is now spent on handling various administrative tasks that may only be indirectly associated with patient care and have high potential for being fully automated (9). For example, AI can be used to manage patient records, schedule appointments, and even assist with billing and insurance claims.
Current General Practice AI tools
One study reviewed the administrative task with augmented artificial intelligence compared to diagnostic artificial intelligence initiatives. The results showed that administrative tasks in general practice have relevant use cases suitable for academic research and high potential for being fully automated by data-driven methods of AI, yet the current quantity and use of cutting-edge machine learning methods (e.g., deep learning using artificial neural networks), when compared to those applied in diagnostic support, appear lacking (10). Increasing the integration of artificial intelligence and automation of administration can help reduce the workload for healthcare professionals, which can lead to better patient care and improved outcomes.
4. Virtual Assistance
Another application of augmented AI is in the development of virtual assistants, which can provide patients with 24/7 access to healthcare information and advice. AI–driven chatbots (AI chatbots) are conversational agents that mimic human interaction through written, oral, and visual forms of communication with a user (11).
Lifestyle Impact Using AI Chatbots
One study reviewed the effectiveness of AI chatbots on various lifestyle changes. This review found that AI chatbots were efficacious in promoting healthy lifestyles, including physical exercise and diet (6/15, 40%), smoking cessation (4/15, 27%), treatment or medication adherence (2/15, 13%), and reduction in substance misuse (1/15, 7%) (11). However, the outcomes of this review need to be interpreted with caution because most of the included studies had a moderate risk of internal validity, given that the AI chatbot intervention domain is at a nascent stage. Virtual assistants can help reduce the workload for healthcare professionals and improve patient access to care.
5. Drug Discovery
Genomics and AI
AI is being used in drug discovery to identify potential drug targets and develop new treatments. AI can recognize hit and lead compounds and provide a quicker validation of the drug target and optimization of the drug structure design (12,13). AI can be used to analyze large amounts of data from genetic and molecular studies to identify potential drug targets, and to predict the efficacy and side effects of new drugs. The process of discovering and developing a drug can take over a decade and costs US$2.8 billion on average. Even then, nine out of ten therapeutic molecules fail Phase II clinical trials and regulatory approval (14,15). Different AI-based tools can be used to predict physicochemical properties. For example, ML uses large data sets produced during compound optimization done previously to train the program (16). Algorithms for drug design include molecular descriptors, such as SMILES strings, potential energy measurements, electron density around the molecule, and coordinates of atoms in 3D, to generate feasible molecules via DNN and thereby predict its properties (17).
Proteomics and AI
The design is in accordance with the chemical environment of the target protein site, thus helping to predict the effect of a compound on the target along with safety considerations before their synthesis or production (18). The AI tool, AlphaFold, which is based on DNNs, was used to analyze the distance between the adjacent amino acids and the corresponding angles of the peptide bonds to predict the 3D target protein structure and demonstrated excellent results by correctly predicting 25 out of 43 structures (19).
Drug Discovery
This approach can help accelerate the drug development process and bring new treatments to market more quickly. The current healthcare sector is facing several complex challenges, such as the increased cost of drugs and therapies, and society needs specific significant changes in this area. With the inclusion of AI in the manufacturing of pharmaceutical products, personalized medications with the desired dose, release parameters, and other required aspects can be manufactured according to individual patient need (20).
Conclusion:
Despite the many benefits of augmented AI in healthcare, there are also potential challenges that must be addressed. For example, there is a risk of biased algorithms if the data used to train the AI system is not representative of the entire population. Additionally, there are concerns around the privacy and security of patient data, as well as the potential for job displacement as automation becomes more widespread. However, with careful development and implementation, the potential benefits of augmented AI in healthcare are vast.
References:
1. Dias R, Torkamani A. Artificial intelligence in clinical and genomic diagnostics. Genome Med. 2019;11(1):70. Published 2019 Nov 19. doi:10.1186/s13073-019-0689-8
2. Lehman, C. D. et al. National performance benchmarks for modern screening digital mammography: update from the Breast Cancer Surveillance Consortium. Radiology 283, 49–58 (2017).
3. McKinney SM, Sieniek M, Godbole V, et al. International evaluation of an AI system for breast cancer screening [published correction appears in Nature. 2020 Oct;586(7829):E19]. Nature. 2020;577(7788):89-94. doi:10.1038/s41586-019-1799-6
4. Hoffman RM, Atallah RP, Struble RD, Badgett RG. Lung Cancer Screening with Low-Dose CT: a Meta-Analysis. J Gen Intern Med. 2020;35(10):3015-3025. doi:10.1007/s11606-020-05951-7
5. Roth, H.R. et al. (2014). A New 2.5D Representation for Lymph Node Detection Using Random Sets of Deep Convolutional Neural Network Observations. In: Golland, P., Hata, N., Barillot, C., Hornegger, J., Howe, R. (eds) Medical Image Computing and Computer-Assisted Intervention – MICCAI 2014. MICCAI 2014. Lecture Notes in Computer Science, vol 8673. Springer, Cham. https://doi.org/10.1007/978-3-319-10404-1_65
6. Awada H, Durmaz A, Gurnari C, et al. Machine learning integrates genomic signatures for subclassification beyond primary and secondary acute myeloid leukemia [published correction appears in Blood. 2022 Mar 3;139(9):1424-1425]. Blood. 2021;138(19):1885-1895. doi:10.1182/blood.2020010603
7. Qin Y, Wu J, Xiao W, Wang K, Huang A, Liu B, Yu J, Li C, Yu F, Ren Z. Machine Learning Models for Data-Driven Prediction of Diabetes by Lifestyle Type. Int J Environ Res Public Health. 2022 Nov 15;19(22):15027. doi: 10.3390/ijerph192215027. PMID: 36429751; PMCID: PMC9690067.
8. Pedersen KM, Andersen JS, Søndergaard J. General Practice and Primary Health Care in Denmark. J Am Board Fam Med. 2012;25:S34. doi: 10.3122/jabfm.2012.02.110216.
9. Lin SY, Mahoney MR, Sinsky CA. Ten ways artificial intelligence will transform primary care. J Gen Intern Med. 2019;34(8):1626–1630. doi: 10.1007/s11606-019-05035-1.
10. Sørensen NL, Bemman B, Jensen MB, Moeslund TB, Thomsen JL. Machine learning in general practice: scoping review of administrative task support and automation. BMC Prim Care. 2023 Jan 14;24(1):14. doi: 10.1186/s12875-023-01969-y. PMID: 36641467; PMCID: PMC9840326.
11. Aggarwal A, Tam CC, Wu D, Li X, Qiao S. Artificial Intelligence-Based Chatbots for Promoting Health Behavioral Changes: Systematic Review. J Med Internet Res. 2023 Feb 24;25:e40789. doi: 10.2196/40789. PMID: 36826990; PMCID: PMC10007007.
12. Mak K.-K., Pichika M.R. Artificial intelligence in drug development: present status and future prospects. Drug Discovery Today. 2019;24:773–780.
13. Sellwood M.A. Artificial intelligence in drug discovery. Fut. Sci. 2018;10:2025–2028.\
14. Álvarez-Machancoses Ó, Fernández-Martínez J.L. Using artificial intelligence methods to speed up drug discovery. Expert Opin. Drug Discovery. 2019;14:769–777.
15. Fleming N. How artificial intelligence is changing drug discovery. Nature. 2018;557 S55–S55.
16. Yang X. Concepts of artificial intelligence for computer-assisted drug discovery. Chem. Rev. 2019;119:10520–10594.
17. Hessler G., Baringhaus K.-H. Artificial intelligence in drug design. Molecules. 2018;23:2520.
18. Wan F., Zeng J. Deep learning with feature embedding for compound–protein interaction prediction. bioRxiv. 2016;2016
19. Paul D, Sanap G, Shenoy S, Kalyane D, Kalia K, Tekade RK. Artificial intelligence in drug discovery and development. Drug Discov Today. 2021 Jan;26(1):80-93. doi: 10.1016/j.drudis.2020.10.010. Epub 2020 Oct 21. PMID: 33099022; PMCID: PMC7577280.
20. Rantanen J., Khinast J. The future of pharmaceutical manufacturing sciences. J. Pharm. Sci. 2015;104:3612–3638.
Acknowledgements:
The authors would like to acknowledge the contributions of the medical professionals, researchers, and AI developers working collaboratively to improve patient outcomes and safety. The advancements in the field of AI and ML algorithms have the potential to revolutionize patient care, and continued efforts in research and development will help to address the challenges and limitations currently faced in the implementation of these technologies.
Conflicts of Interest:
The authors declare no conflicts of interest.
Funding:
No funding was received for this study.
Author Contributions:
All authors contributed equally to the conception and design of the study, data collection and analysis, and manuscript preparation. All authors have read and approved the final manuscript.
Ethics Approval and Consent to Participate:
Not applicable.
Consent for Publication:
Not applicable.
Availability of Data and Materials:
Not applicable.
Competing Interests:
The authors declare that they have no competing interests.
