Artificial Intelligence in Healthcare – Data Science Central

Introduction

Artificial intelligence refers to simulating the behavior of humans, so that machines can be programmed to perform intelligent behavior and mimic human actions. It is a branch of computer science dealing with building smart machines which can perform actions, typically needing human intelligence. With the availability of huge data, faster computation power, and technology advancement in machine learning and deep learning is providing a paradigm shift in across all the sectors. Artificial Intelligence (AI) in healthcare leverages complex algorithms to emulate human behavior in the data exploration, analysis and training the models, and comprehension of complicated medical and healthcare data.  In this article, we will review the key applications of artificial intelligence in the healthcare sector.

Abstract

Artificial intelligence (AI) has made significant progress in the recent years and is poised to transform the healthcare sector. Deep learning algorithms can deal with increasing amounts of data provided by wearables, smartphones, and other mobile monitoring sensors in different areas of medicine (Briganti & Le Moine, 2020). This article provides a perspective on how medical field can leverage AI in the future. It includes predictive modeling, and concepts such as feature selection, common algorithms used in the supervised learning and the selected application in the medical field. Also, it includes how deep learning, unsupervised learning techniques can be used to improvise patient outcomes.

Significance of AI in healthcare

Correct diagnosis of the diseases by a human needs years of medical study, and still the manual diagnosis is an arduous and very time consuming process. Hence, the demand for experts is ever rising, which puts huge strain on the healthcare professionals and can also lead to delay in the diagnosis of life saving patients. Deep Learning, Machine Learning algorithms have made huge advancement which can make the diagnosis much faster, cheaper and more accessible. Machine learning algorithms can learn from vast available data and accurately classify the patterns in fraction of seconds. Some of the common applications –

• Lung cancer detection from the CT scans
• Diabetic retinopathy indicators from the eye images
• Skin Lesions classification from the skin images
• Analyzing the risk of cardiac arrest from cardiac MRI images

AI is relevant to many healthcare areas including visually-orientated specialties such as radiology, pathology, ophthalmology, and dermatology due to the availability of large digital datasets. Deep learning algorithms leverage these datasets to train themselves and perform a specific tasks e.g. identifying a lesion in an image (Kulkarni et al., 2020). Precision medicine has the potential to improve the traditional symptom-driven practice of medicine by intelligently integrating multi-omics profiles with clinical, imaging, epidemiological and demographic details to allow a wide range of earlier interventions for advanced diagnostics and tailoring better and economical personalized treatment. Below figure depicts the role of artificial intelligence in traditional healthcare data analytics, and in precision medicine(Ahmed et al., 2020).

Machine learning algorithms can be broadly categorized under supervised, unsupervised and reinforcement learning. While supervised learning focus on classification / regression based on intelligence from historical data, however unsupervised learning focus on identifying hidden patterns and relationships from unlabeled data. Reinforcement learning is based on learning the behavior through trial and error from input data, while trying to optimize the outcome.

Above figure depicts the typical components of machine learning cycle. It starts with data preparation and cleaning and applying transformations, normalizations or encoding, which is extremely critical for the performance of machine learning models. The next step involves selecting the right set of features to avoid overfitting or underfitting of the machine learning models. It can also include feature engineering, which leverage domain knowledge to create new features for improvising the machine learning models. The subsequent stages involves building machine learning models, training, optimizing, validating and selecting machine learning models to solve a problem (Waring et al., 2020).

The central promise of machine learning is to incorporate data from a variety of sources (clinical measurements and observations, biological –omics, experimental results, environmental information, wearable devices) into sensible models for describing and predicting human disease. The typical machine learning workflow begins with data acquisition, proceeds to feature engineering and then to algorithm selection and model development, and finally results in model evaluation and application. Below figure provides the overview of a typical machine learning workflow in the healthcare industry (Johnson et al., 2018) –

Recent Applications of Artificial Intelligence in Healthcare

With the emergence of massive compute power and data generated in the healthcare systems, it has provided good emergence of new AI applications, which also include faster development and trails of Covid-19 vaccine. Below are two recent applications, which are accurate and clinically relevant to benefit both the patients and the doctors by making diagnosis more straightforward.

The first of these algorithms is one of the multiple existing examples of an algorithm called DLAD (Deep Learning based Automatic Detection) to analyze chest radiographs and detect abnormal cell growth, such as potential cancers. The algorithm’s performance was compared to multiple physician’s detection abilities on the same images and outperformed 17 of 18 doctors. The second of these algorithms, LYNA (Lymph Node Assistant), to identify metastatic breast cancer tumors from lymph node biopsies. This isn’t the first application of AI to attempt histology analysis, but interestingly this algorithm could identify suspicious regions undistinguishable to the human eye in the biopsy samples given. LYNA was tested on two datasets and was shown to accurately classify a sample as cancerous or noncancerous correctly 99% of the time. 

Applications of AI algorithm (DLAD) in medicine. The left panel shows the image fed into an algorithm. The right panel shows a region of potentially dangerous cells, as identified by an algorithm, that a physician should look at more closely.

Both LYNA and DLAD serve as prime examples of algorithms that complement physicians’ classifications of healthy and diseased samples by showing doctors salient features of images(Greenfield, n.d.).

Conclusion

The advancements of new techniques in artificial intelligence in clinical practice are significantly helping the patients and the healthcare professionals in accurately and faster diagnosis of the diseases, developing drugs and providing personalized treatments. It is a promising area for development which is rapidly evolving along with other modern areas genomics, precision medicines and teleconsultation. While the scientific research can help in faster development of new solutions to help the healthcare, more rigorous policies should be in place to ensure ethical usage from the evolution of the medicines. It is also significant for the physicians to be aware of the recent advancements in AI, which is going to transform the healthcare in the future.

Bibliography:

Ahmed, Z., Mohamed, K., Zeeshan, S., & Dong, X. (2020). Artificial intelligence with multi-functional machine learning platform development for better healthcare and precision medicine. Database, 2020. 

Briganti, G., & Le Moine, O. (2020). Artificial Intelligence in Medicine: Today and Tomorrow. Frontiers in Medicine, 7, 27.

Greenfield, D. (n.d.). Greenfield D. Artificial intelligence in medicine: applications, implications, and limitations. Published 2019. Accessed 8 Jan 2020. 

Johnson, K. W., Torres Soto, J., Glicksberg, B. S., Shameer, K., Miotto, R., Ali, M., Ashley, E., & Dudley, J. T. (2018). Artificial Intelligence in Cardiology. Journal of the American College of Cardiology, 71(23), 2668–2679.

Kulkarni, S., Seneviratne, N., Baig, M. S., & Khan, A. H. A. (2020). Artificial Intelligence in Medicine: Where Are We Now? In Academic Radiology (Vol. 27, Issue 1, pp. 62–70). Elsevier USA. Waring, J., Lindvall, C., & Umeton, R. (2020). Automated machine learning: Review of the state-of-the-art and opportunities for healthcare. Artificial Intelligence in Medicine, 104, 101822.