1. Introduction
An information system is a combination of information technology and the service to patients of the clinic, according to their needs. Medical clinic software applications serve as an essential tool that is revolutionizing healthcare for both providers and patients [
1,
2]. The multifaceted benefits it offers include streamlined access to medical information, improved time efficiency, patient convenience, and a host of features that contribute to proactive healthcare management.
One of the key benefits of a medical clinic app is its ability to optimize the time of both patients and healthcare providers. Patients can speed up the appointment scheduling process, view available time slots, and receive timely reminders, thereby reducing clinic wait times. This time efficiency is not only convenient for patients, but also contributes to the overall operational efficiency of the healthcare facility.
Furthermore, software applications extend the healthcare sphere beyond the confines of physical clinics [
3]. Patients can connect remotely with healthcare providers through telemedicine services, enabling virtual consultations. This is proving particularly valuable for routine check-ups, follow-up appointments and discussion of laboratory results, removing the need for unnecessary in-person visits and improving accessibility to healthcare services.
Patient convenience is further enhanced by features such as prescription refills and secure messaging with healthcare professionals [
4]. Patients can request prescription refills through the application, saving them from having to visit the clinic for routine medication needs. Secure messaging promotes effective communication by allowing patients to seek clarification, ask questions and receive guidance from healthcare professionals, strengthening the doctor–patient relationship.
Such a software product can also serve as a comprehensive health management tool [
5]. Patients can monitor and track various health parameters, such as blood pressure, glucose levels or physical activity, contributing to a more holistic approach to healthcare. Automated alerts and reminders for medication schedules, upcoming appointments or preventive checkups further enhance the app’s functionality in promoting health and well-being.
Including a blog section in the app adds an educational dimension. Patients can access health-related articles, tips and updates that enrich their knowledge about specific conditions, preventative measures, and general wellness practices [
6]. This educational content not only informs patients, but also encourages them to actively participate in their health journey. For healthcare providers, the app streamlines administrative tasks, from appointment management to electronic record keeping. Efficient document handling and the ability to maintain patient records digitally contribute to a more organized and efficient healthcare workflow.
In conclusion, medical clinic software has emerged as a transformative tool that not only provides easy access to information but also redefines the patient experience by saving time, promoting convenience and active participation in healthcare management. Its multifaceted features contribute to a more patient-centric and efficient healthcare ecosystem [
7,
8].
A single mobile app for medical clinics can offer several benefits to both healthcare providers and patients. Here are some reasons why such an app could be valuable:
Access to medical information: patients can access their medical records, details of examinations and results via the mobile app.
Time Efficiency: Patients can save time by using the app to schedule appointments, view available time slots and receive reminders.
Patient convenience: the mobile app allows patients to manage their healthcare from the comfort of their own home.
Remote consultations: some mobile apps enable telemedicine services by allowing patients to consult with medical professionals remotely.
Health monitoring and alerts: Patients can use the app to track and monitor their health parameters. such as blood pressure, glucose levels or physical activity.
Educational content.
Improved communication.
Appointment reminders and notifications.
In summary, a mobile app for medical clinics can improve the overall patient experience by providing easy access to information, improving communication, saving time, and encouraging proactive healthcare management [
9].
3. Artificial Intelligence Process
Artificial intelligence process consists of six phases.
Study of a suitable dataset;
Data analysis and visualization of various trends;
Normalization to the data;
Processing;
Training the models with artificial intelligence;
Exporting the model so it can be integrated into the Flask Web API.
3.1. Features for Hearth Disease Dataset [10]
This is a multivariate type of dataset, providing or including a variety of separate mathematical or statistical variables, i.e., multivariate numerical data analysis. It consists of 14 characteristics: age, gender, type of chest pain, resting blood pressure, serum cholesterol, fasting blood sugar, resting electrocardiographic results, maximal heart rate achieved, exercise-induced angina, old peak—ST depression, exercise-induced vs. rest, peak exercise ST-segment slope, large vessel number, and thalassemia. This database includes 76 attributes, but all published studies refer to the use of a subset of 14 of them.
The Cleveland database is the only one used by ML researchers to date. One of the main tasks of this dataset is to predict based on the given attributes of a patient whether that particular person has a heart disease or not, and another is the experimental task of diagnosing and locate various insights from this dataset that could help to understand the problem better (
Table 1). The dataset consists of about 300 records.
The features cp, thal, slope are of type Category and have values that have different weights for assessing whether a patient has heart problems. Columns need to be added for:
Cp—cp_0, cp_1, cp_2, cp_3;
thal—thal_0, thal_1, thal_2;
slope—slope_0, slope_1, slope_2.
After adding the new columns, the cp, thal and slope columns are removed.
The dataset is split into 80% for training data and 20% for test data, and 10% of the data are scrambled. This action was taken in order not to overfit the model, i.e., so that it recognizes only the training data.
At the end of the training experiments performed, results were obtained with an accuracy of 85.00%.
3.2. Features for Diabetes Dataset [11]
This dataset is originally from the National Institute of Diabetes and Digestive and Kidney Diseases. The purpose of the dataset is to diagnostically predict whether or not a patient has diabetes based on certain diagnostic measurements included in the dataset (
Table 2). Several constraints were placed on the selection of these instances from a larger database. Notably, all patients here are women at least 21 years of age of Pima Indian heritage.
The dataset consists of about 800 records.
In this dataset, there is no need for data normalization because we have medical values that do not allow this operation to be performed.
When interpreting features in some cases, especially when the features represent specific medical measures, the original values should be preserved in order to easily interpret the results. When the features represent physiological measurements, such as blood pressure or insulin levels, leaving the original values may be medically preferred.
As regard feature scaling, if the features are on appropriate scales, for example if they are measured in the same units, and the scale differences are not significant, then normalization may not be so critical.
The dataset is split into 80% for training data and 20% for test data, and 42% of the data are scrambled. This action was taken in order not to overfit the model, i.e., so that it recognizes only the training data.
At the end of the training experiments performed, results were obtained with an accuracy of 77.92%.
3.3. Artificial Intelligence Algorithms for both AI Models
Logistic regression is a commonly used algorithm in medical applications, including heart disease detection, for these reasons:
Output is the probability of an event—logistic regression produces an output in the interval [0, 1], which can be interpreted as the probability of having heart disease. This is useful in medical applications where the interest is in predicting the probability of a particular event.
Interpretation of coefficients: The coefficients in logistic regression can be interpreted as the influence of each characteristic on the logarithm of the odds ratio. This gives a clear picture of which factors are associated with greater or lesser chances of heart disease.
Relative ease of training: Logistic regression is generally faster to train than more complex models and requires fewer training data. This can be an advantage, especially when you have limited medical data.
Dealing with highly correlated factors: Logistic regression is relatively robust to the problem of multicollinearity (highly correlated predictors). This is important because medical data often have multiple interrelated characteristics.
Interpretability: Logistic regression results are easy to interpret and explain, which can be important in the medical field where clear communication with doctors and patients is needed.
Ability to work with linear and non-linear interactions: Logistic regression can capture linear and non-collinear interactions between characteristics. This is useful because complex interactions between different factors are often encountered in medical data.
Robustness to the problem of outliers: Logistic regression is relatively robust to outliers in the data, which is important in the medical context, where data can often be affected by outliers.
Relatively easy to train: Logistic regression can be trained quickly, especially if the data are relatively large. This makes the algorithm suitable for use with medical data, which is often huge and diverse.
AI Models settings:
The model was trained with the scikit learn library, sklearn.linear_model.LogisticRegression
Features: Penalty: L2 regularization (Ridge).
where
J(
θ)—
Target function;
λ—tuning parameter specifying the strength of regularization;
—the sum of the squares of the parameter values.
LBFGS is an optimization algorithm that is used to find the minimum of a function. LBFGS (Limited-memory Broyden–Fletcher–Goldfarb–Shanno) is the most widely used version of the Broyden–Fletcher–Goldfarb–Shanno method. It is a method for optimizing functions of multiple variables where no information about the second derivatives of the function is required.
5. Conclusions
The current development proposes an information system for a medical clinic service. It provides information security against known cyber-attack methods. The software allows patients to retrieve medical prescriptions, share or download electronic medical prescriptions, preview medical examinations, blog, and schedule appointments.
Despite the available functionalities, the system could be improved and developed with the following future developments:
Integrate an option to use supplemental health insurance when booking an appointment;
Notify patient of upcoming appointments with doctor;
Digital authorization for medical prescription renewal in order to continue treatment;
Remote consultations;
Physical activity monitoring;
- ○
Health parameters:
- ○
Record and store blood pressure values;
- ○
Recording and storage of blood oxygen values;
- ○
Record and store blood glucose values;
- ○
Integration of specialized glucose monitoring sensor (Continuous Glucose Monitors, CGM), for blood glucose levels.
Keeping track of these patient health indicators will help collect data on their health. This data will be used to analyze the patients’ condition and to structure datasets. These datasets will be used to train various AI modules that will help doctors in diagnosing patients.