Application of AI in the medical device field
This article originally appeared on the PQE Group blog and was written by Monica Magnardini, Medical Device Compliance Specialist and Biomedical Engineer at PQE Group.
July 17, 2024
| BioBuzz has been connecting life science talent since 2009. We've built a vast community in the Mid-Atlantic and have readers across the country, from Massachusetts to Florida, New York to California. Our next chapter is building a community of our own. Human Resources Logistics A model to support employers Finding and hiring life science talent. learn more. |
In this article, we delve into the multifaceted world of artificial intelligence, tracing its origins from philosophical speculations to a landmark conference at Dartmouth College in 1956. Defined as the ability of a system to simulate human intelligence, AI has evolved significantly, especially through the fields of machine learning and deep learning. The healthcare industry has emerged as the biggest beneficiary of AI's capabilities in a variety of applications, from disease detection to personalized treatment recommendations.
Real-world examples show how AI is transforming medical devices, digital health technology, and even hearing aids, promising a future where healthcare is not only more efficient, but also more personalized and accessible.
introduction
The term “artificial intelligence” (AI) is notoriously difficult to define. Often, AI is used to mean things that computers have difficulty doing (such as understanding natural language) rather than things that they can do reasonably well (such as accounting).
But what is artificial intelligence?
According to ISO/IEC 2382:2015, artificial intelligence is defined as the ability of a system to perform tasks or develop data processing systems that perform functions normally associated with human intelligence.
In other words, artificial intelligence (AI) is a specific type of technology that enables computers and machines to simulate human intelligence and problem-solving abilities.
Two fields – machine learning and deep learning – make it possible to develop AI algorithms that are modeled after the decision-making processes of the human brain and can learn over time, becoming more and more accurate at making predictions.
History of AI
So where does the term “artificial intelligence” come from?
At a conference held at Dartmouth College in the summer of 1956, several scientists discussed ways to make machines simulate intelligence. It was McCarthy who first coined the term “artificial intelligence.”
However, the history of artificial intelligence began in ancient times, with myths, stories and rumors about artificial creatures that were given intelligence and consciousness by skilled craftsmen. The seeds of modern AI were sown by philosophers who tried to explain human thought processes as the mechanical manipulation of symbols. This research culminated in the 1940s with the invention of the programmable digital computer, a machine based on the abstract nature of mathematical reasoning. This device and the ideas behind it inspired a handful of scientists to start seriously discussing the possibility of building an electronic brain. Indeed, the first artificial neuron was designed by McCulloch and Pitts in 1943.
Since that conference in 1956, artificial intelligence (AI) has been used in various forms and degrees to develop and advance a wide range of sectors, including healthcare, manufacturing, banking and financial markets, education, supply chains, retail, and e-commerce.
AI in Healthcare
In the healthcare industry, AI-based medical devices can automate tasks and integrate data from multiple sources to pinpoint trends, process and analyze information from wearable sensors to identify the onset of diseases and medical conditions, predict which patients are at high risk of disease, complications, or poor outcomes based on medical records, and support research by evaluating large amounts of data and monitoring the effectiveness of treatments.
Machine learning techniques evaluate structured data such as images, genetic and electrophysiological data to help outline patient characteristics and predict disease, while deep learning techniques are used for more complex data taken from medical datasets.
Medical sensors play an important role in the diagnostic field as they can convert biomedical parameters into easily measurable signals, making diagnostic equipment more effective and safer. A wide range of medical sensors are being developed for disease monitoring and diagnosis, including biomedical markers that can be used on or inside the body. These types of sensors are actively used, for example, to detect cancer.
Meanwhile, as technology advances, software incorporating artificial intelligence (AI) – particularly the subset of AI called machine learning (ML) – is becoming an essential component of an increasing number of medical devices.
The greatest benefit of AI/ML in software is its ability to learn and improve performance from real-world usage and experience. The ability of AI/ML software to learn from real-world feedback (training) and improve performance (adaptation) gives these technologies a unique place in the world of Software as a Medical Device (SaMD) and the rapidly expanding field of research and development.
Case studies
AI/ML has the potential to generate new and important insights from the vast amounts of data generated during daily healthcare delivery. Digital health technologies are playing an increasingly important role in many aspects of our health and daily lives, and AI/ML is driving important advances in this field.
An example of AI/ML software could be an AI/ML application designed for ICU patients that receives electrocardiogram, blood pressure, and pulse oximetry signals from a primary patient monitor. In this example, the physiological signals are processed and analyzed to identify patterns that occur at the onset of physiological instability. If physiological instability is detected, an audible alarm signal is generated to indicate the need for immediate clinical attention to prevent potential harm to the patient.
Another example is a mobile app that aims to provide a risk assessment of skin lesions using the mobile device's camera and flashlight. It is an AI algorithm that performs the risk assessment. Based on the assessment results, the user is advised to undergo further examination by a dermatologist to get an accurate medical diagnosis. To avoid any misinterpretation on the results screen, the algorithm draws a box around the risk-assessed lesion, colored according to the assigned risk, thus indicating exactly which lesion the risk assessment was calculated for. If there are multiple lesions, multiple boxes are created with the corresponding colors. If the algorithm fails to detect a lesion, no box is displayed on the results screen and the user cannot assess it.
Continuous glucose monitoring (CGM) and self-management mobile apps have been used in recent times to lead to the digital transformation of diabetes care. Insulin bolus calculators were developed to assist in adjusting insulin doses and are now included in most modern insulin pumps and some blood glucose meters on the market. By utilizing continuous glucose monitoring (CGM), run-to-run control, and artificial intelligence, devices have been developed that overcome this barrier. The algorithm implemented in the mobile app communicates in real time with the continuous glucose sensor and appears to require the user to manually input various information to calculate the recommended insulin dose tailored to the individual and their current situation. The algorithm adopts recommendations based on the results of past recommended insulin doses and user behavior.
In the hearing aid sector, the first hearing aids have been introduced to the market that use real-time machine learning to allow the end user to make adjustments based on their preferences and intents in different environments. By utilizing a distributed computing approach and a smartphone connected to the hearing aid, a live machine learning application can be incorporated. Through a simple interface, the algorithms can automatically learn and meet the end user's preferences and intents. In other words, machine learning allows the equalizer settings to be adjusted without changing the programming that the hearing healthcare professional built into the fitting. The permanent programming of the hearing aid remains unchanged, but the end user can easily adjust the acoustic settings in real time to suit their specific real-time listening intents.
Conclusion
In the ever-evolving field of medical technology, artificial intelligence stands as a beacon of innovation, offering solutions that were once unimaginable. From real-time monitoring to personalized treatment plans, AI is not only increasing the efficiency of medical interventions but also empowering patients to take control of their health like never before. Looking to the future, the convergence of AI and healthcare promises a world where diseases are detected earlier, treatments are more effective, and individuals can live healthier lives.
- About the Author
- Latest Posts
BioBuzz is a community-driven, experience-focused biotech and life science media and events company. BioBuzz connects you to the latest local news, industry experts, jobs, events, and resources to help grow your business and career. Their weekly newsletter is the go-to source for industry updates for thousands of subscribers across the BioHealth Capital Region and Greater Philadelphia area.
