Predicting Heart Failure. Группа авторов

of the test gives information about heart health and helps to gather information about heart arrhythmias. Echocardiography outputs have been used with both machine learning algorithms and deep learning algorithms [8].

      Computed imaging (tomography) is used to detect other diseases as well as in HF. It creates three-dimensional (3D) images that can show the obstructions in our main vessels with the help of a computer system. Through computed tomography, testing can be performed, especially for aortic disease.

      The exercise test works with skin-attached electrodes and a monitor recording a person’s heart function while walking on a treadmill. Many aspects of heart function can be checked, including heart rate, breathing, blood pressure, ECG, and how tired someone is while exercising. This test is particularly helpful in diagnosing CAD and is also used to predict heart attack risk. It helps diagnose the possible cause of symptoms such as chest pain (angina).

The last procedure on this subject is known as the thallium stress test. The thallium test is similar to a routine exercise stress test. The difference is that the radioactive thallium material injected into the patient’s blood is photographed by special gamma ray cameras when the patient is at the maximum exercise level. Thus, the extent of coronary artery occlusion can be determined and the extent of damage from heart attack can be determined [9].

      1.4 Computer-Aided Diagnosis

The number of diseases and patients is increasing daily due to food-related problems, a sedentary lifestyle, and many similar reasons. Factors such as the complexity of the diseases, the increasing number of patients, and the aging of the world population have led to new searches in the field of health. Diseases are not old diseases, and their solution methods cannot be old methods. Being aware of this situation, health professionals and technology workers have joined hands and developed solutions for the health sector. In this context, many studies using technology in the field of health have been carried out [10]. Notably, using internet of things (IoT) sensor technologies, human movements have been monitored and abnormal situations detected [11].

The first solution that comes to mind regarding data collected with the help of IoT sensors is alarm generation resulting from abnormalities in the collected data, and sometimes making diagnoses and patient interventions the patient [12]. Abnormalities in patient movements or blood values ​​are usually associated with a disease. For example, irregular tissue growths are related to cancer, and irregular heartbeats, heart diseases, and balance problems in the body are related to nerve diseases. IoT is a network where physical objects communicate with each other and with central systems. The use of IoT devices in disease diagnosis is carried out with the help of wearable technologies or additional components that make it possible to generate alarms about the patient with the help of systems that collect information, especially heart rate information, from the patient and transmit it to the center. This is carried out according to a method called anomaly detection in the name of alarm generation. An alarm is generated in case of an abnormality by comparing the last state of the patient with their recorded normal state. Thus, an approach similar to anomaly detection in networks provides the background for diagnosis based on IoT systems.

HF is a complex disease that occurs due to many different causes. The complexity of the disease makes its correct diagnosis a challenging problem for doctors. In addition, heart diseases, and especially heart attacks, are diseases that frequently affect people over the age of 65 who choose to live at home rather than a health center, and thus it is not possible to keep the patient’s condition under control. Due to the increase in cardiac diseases and the inability of patients to be under continuous control, studies to detect abnormalities in the heart before the onset of a crisis should be carried out earlier [13]. Early diagnosis is important as it affects the early intervention for the patient.

In the diagnosis of HF, the risk of HF is estimated based primarily on the patient’s clinical history, current symptoms, physical examination, and ECG. If all the measurements mentioned are normal, there is probably no risk, otherwise more detailed analysis should be done for diagnosis. Performing analyses allows specialists to identify patients who need echocardiography [14]. Diagnosis can be made with clinical methods, and decision support systems based on machine learning algorithms. From a machine learning perspective, HF diagnosis is a binary classification problem, a type of classification in which samples are assigned to one of two groups. For the diagnosis of HF, there would be two classes: “heart failure” and “no heart failure.” Thanks to machine learning, it can be understood whether the person being controlled has HF.

      1.4.1 Clinical Decision Support Systems

      There is a need to develop a system in order to collect patient data and make diagnoses based on them. These systems are generally called decision support systems. Decision support systems work by analyzing the data collected from the terminal units in accordance with the needs in the center to support decision making and presenting the results. Decision support systems can sometimes be developed for the field of health and sometimes for another field. Regardless of the field, the task of decision support systems is to infer with the help of the collected data and the rules at hand.

Decision support systems developed specifically for the health field are often referred to as clinical decision support systems. Clinical decision support systems sometimes help experts diagnose disease and recommend treatment methods based on health information. The concept of computer-aided clinical decision making came to the fore many years ago [15]. The working principle of clinical decision support systems is simply as follows: information such as blood values and enzyme values of the patient is obtained, this information is evaluated according to the knowledge base of the health institution, and an inference is made about the patient. When a direct relationship can be established between the inputs and the final decision during inference, this is known as an open box decision support system; when the correct relationship cannot be established between the inputs and the final decision, it is called a closed box decision support system [16].

      A decision support system is a software that helps decision making and works based on the ruleset. The rule set can be in the form of simple IF-THEN rules or a more complex structure. Rule sets can be created with the help of expert knowledge or with models that learn from data. Comparing the event that occurred with the rules in the rule set determines which rule triggered the situation. Thus, whichever rule is triggered by the situation to produce the result, the output of the decision support system will be consistent.

      Decision support systems are systems that help to decide about a situation. A system that will decide on HF takes the patient’s measurements as input and provides information about the risk of HF as output. The system’s inputs are sometimes the information coming from the IoT devices and sometimes the data that has been entered manually. For example, based on heart rate and blood values, it will be possible to predict whether there will be HF. At this point, we come across systems based on machine learning and knowledge. In machine learning based systems, data classification is carried out with the help of models trained with historical data. As a result of the classification, there might or might not be an indicated risk of HF.

      1.5 Diagnosis with Artificial Intelligence Methods

One of the techniques that can be used to diagnose of HF is artificial intelligence and machine learning.

      1.5.1 Introduction to Artificial Intelligence

      Artificial intelligence is the name of the general discipline that combines software and hardware studies. It includes designing systems that act like a human, learning and making decisions like humans. Scientists have been curious about artificial intelligence since the 1950s and there is increasing interest in it. It has become an even more important discipline with the developments in information technologies. The focus on developing systems that exhibit human-like behavior has already attained this goal in some areas. Examining the recent history of artificial intelligence shows that machine learning algorithms and then deep learning algorithms have led to striking developments. For example, thanks to natural language processing, systems that produce