Deep learning has the potential to change the future of medical science with successful experimental results and widespread applications. Nowadays, the use of artificial intelligence has become increasingly common and is used in various fields such as cancer diagnosis. Deep learning also enables computer vision, imaging, and more accurate medical diagnosis.
Therefore, it is not surprising that a report from Report Linker has indicated that the artificial intelligence market in the medical industry is expected to reach 36 billion dollars in 2025 from 1.2 billion dollars in 2018!! In this article, we explore the potential of deep learning in the healthcare and medical industry and its many applications in this field.
Table of Contents
Deep learning: the future of medical science
As deep learning and artificial intelligence gain popularity in the industry, the question arises as to how they will affect our lives in the next few years. In the medical field, although we have recorded a large amount of patient data over the past few years, so far deep learning has mostly been used to analyze data from images or text. In addition, deep learning is recently being used to predict a wide range of clinical problems and outcomes. Deep learning has a great future in medicine.
Today’s interest in deep learning in medical science stems from two factors. First, the development of deep learning techniques widely. Especially unsupervised learning methods in areas like Facebook and Google. Second, the dramatic increase in healthcare data
Using deep learning in electronic health records
Electronic health systems store patient data such as demographic information, medical records, and test results. Using deep learning algorithms, these systems improve the rate of correct diagnosis and the time required to diagnose the disease. These algorithms use data stored in electronic health systems to detect health trend patterns and risk factors and draw conclusions based on the identified patterns. Also, researchers can use the data in electronic health systems to create models in deep learning that predict the probability of certain health-related outcomes.
Two ways to use data from electronic health systems
1. Static prediction
Static prediction expresses the probability of an event occurring based on researchers’ datasets from the International Statistical Classification of Diseases and Health Problems. For example, Choi and colleagues tested a model based on electronic health system data, such as medical records and hospital visit rates. Based on this information, the system predicted the probability of heart failure.
2. Prediction based on a set of inputs
Data from electronic health systems are used to make predictions based on a set of inputs. Prediction can be done with each input or with the entire data set. For example, Choi and his colleagues have developed a model by this method. Using artificial neural networks, this model predicts the time of the next visit to the hospital and the reason for the visit.
Application of deep learning in medicine
Deep learning techniques use data stored in electronic health records to address many of the concerns of medical care, such as reducing the rate of misdiagnosis and predicting the outcome of procedures. By processing large amounts of different sources such as medical imaging, artificial neural networks can help doctors analyze information and diagnose several diseases:
- Blood sample analysis
- Examining glucose levels in diabetic patients
- Diagnosis of heart problems
- Image analysis to detect tumors
- Detection of cancer cells and cancer diagnosis
- Diagnosis of arthritis from MRI before the onset of injury
Using deep learning for cancer diagnosis
Oncologists have been using medical imaging methods such as computed tomography, magnetic resonance imaging, and X-rays for years. While these systems have been proven to be effective for many types of cancer; A large number of patients suffer from cancers that cannot be diagnosed with these devices. Neural networks such as convolutional neural networks hold promise for the future of cancer diagnosis. Based on the same medical images, artificial neural networks can detect cancer in the early stages with a lower error rate and provide better results for patients. Recently, scientists have succeeded in training different models of deep learning to diagnose different types of cancer with high accuracy.
Systems based on deep learning make decisions and execute specific commands by imitating human thought patterns and through neural network algorithms.
The neural layers of deep learning systems are not designed and built by engineers; Rather, it is these different data and information that lead to the progress and improvement of the learning process of these algorithms.
In the following, we examine examples of scientists’ research:
- In a study published by Nvidia, a deep learning model was able to reduce breast cancer misdiagnosis by 85%.
- Inspired by his roommate, who was diagnosed with leukemia, Hossam Haick tried to create a device to cure cancer. Based on his plan, a team of scientists trained a neural network model to identify 17 different diseases based on the smell of patients’ breath with 85% accuracy.
- A team of Enlitic researchers introduced a device that surpassed the ability of a group of expert radiologists to detect lung cancer nodules and reached a detection rate 50% more accurate than human detection in experimental conditions.
- Google scientists have developed a convolutional neural network model that detects metastatic breast cancer from pathology images faster and more accurately. This model achieved 99% success.
Deep learning in disease prediction and treatment
In 2006, the hospitalization cost of people who suffered from preventable diseases in the United States reached 30 billion dollars. Half of hospitalized patients suffer from two diseases: heart problems and diabetes. Deep learning can be used to improve the rate of diagnosis and the time required to create pre-awareness. This can drastically reduce the number of hospitalizations.
Some research teams are already working on their solutions to this problem.
Diabetic retinopathy
In developing countries, more than 415 million people suffer from a type of blindness called diabetic retinopathy, which is a complication of diabetes. Deep learning can help prevent this disease. A model of artificial neural networks can work with data taken from retinal imaging, hemorrhage detection, early signs, and indicators of diabetic retinopathy.
Diabetic patients suffer from this complication due to severe changes in blood sugar levels. Meanwhile, diabetic patients can be controlled in terms of glucose levels. A deep learning model can use this data to predict when patients’ blood glucose levels will rise and fall, allowing them to react by eating a high-sugar snack or injecting insulin.
Human immunodeficiency virus (HIV)
More than 36 million people worldwide suffer from the immunodeficiency virus. These people need to receive a daily dose of antiviral drugs to treat their condition. HIV can mutate quickly. Therefore, to continue HIV treatment, we need to change the drugs prescribed to patients. Using a deep learning model called reinforcement learning can help us deal with these types of viruses. In this way, the convolutional model can track many biomarkers using each drug dose and provide the best course of action for continuous treatment.
A team of researchers at the University of Toronto has developed a tool called DeepBind. A convolutional neural network model that takes genomic data and predicts the sequence of DNA and RNA binding proteins. Researchers can use DeepBind to create computer models that show the effects of changes in the DNA sequence. They can use this information to develop more advanced diagnostic tools and drugs.
Privacy issues arising from the use of deep learning in healthcare
Despite the many benefits that the use of electronic health systems brings; There will still be risks involved. The data stored in these systems carry the personal information of patients, which in many cases people prefer to keep this information confidential. Hospitals also store non-medical data, such as patients’ addresses and credit card information; which makes these systems prime targets for attack. Despite the sensitive data stored in electronic health systems and their vulnerability, it is very important to protect it and preserve the privacy of patients.
It is not surprising that shortly, we will see that the average human “life expectancy” will increase by 20 years; And this will not be possible except by artificial intelligence and deep learning techniques.