International Conference on Applied Technologies

Overweight is a serious problem in Peru, and compliance with a healthy and balanced diet is essential for its management. Despite the existence of different types of diets and meal plans, people still find it difficult to comply with the treatment. Also, lack of access to adequate nutritional information and lack of follow-up in implementing a low-calorie diet can demotivate people and reduce the effectiveness of the process. To reduce this problem, a mobile application is presented that allows the control of the caloric intake of food, based on food suggestions using Machine Learning. For this purpose, the Support Vector Machine algorithm is applied to train a model that recommends personalized food to users according to their preferences. This is how the application allows users to easily access personalized food recommendations by analyzing their preferences and caloric needs individually. As a result, it has achieved a Percentage of Caloric Recommendation Satisfaction of 90.5%, a Percentage of Accepted Recommendations of 94.2%, and a Percentage of Usage Satisfaction of 92.6%. These results support its effectiveness and ease of use. It is expected that this proposal will have a positive impact on the fight against overweight in Peru.

The main objective of this research work is to apply Machine Learning to gas sensors, to analyze air quality and transmit the data to a server through the Internet of Things (IoT). The Arduino MEGA 2560 was used, the Ethernet shield module and four gas sensors, MQ2, MQ5, MQ7 and MQ135, were placed; Machine Learning was designed with Artificial Neural Networks (ANN) and greater effectiveness was obtained using the backpropagation training algorithm. The CRISP-DM methodology was used, which contains seven stages, first the problem was identified, in the second stage the data understanding, the third data preparation, in the fourth the Machine Learning with artificial neural networks was designed, the fifth was the modeling, the sixth is the implementation of the model and the seventh is the validation by performing the most common gas tests, resulting in the recognition of the gases in its environment, obtaining a functional system. In conclusion, it was verified and confirmed that Machine Learning with Artificial Neural Networks and the use of the Backpropagation algorithm can automatically detect gases and show the Air Quality Analysis. This research is exclusive for those interested who are beginning their investigation into the world of Machine Learning.

Medical records are a source of valuable information. Processing enormous amounts of data effectively while looking for patterns of interest is made feasible by utilizing artificial intelligence and machine learning techniques. In this research, we apply clustering methods to identify patterns in the health records of women, including age, medical condition, illness, contraceptive methods, and gynecologic features. The methodology used includes data understanding, preprocessing, modeling, and evaluation. For data clustering, three unsupervised algorithms -k-means, DBSCAN, and hierarchical clustering-were applied. To evaluate each technique’s effectiveness, the silhouette metric was used. The experiment results highlight that the optimal silhouette value of 0.73 was achieved with the DBSCAN algorithm by grouping data into 9 clusters. These findings greatly advance our understanding of the most common genital infections and improve our capacity to identify unique patterns within each cluster.

Cataracts are a degenerative disease that causes opacity in the crystalline lens. They represent one of the leading causes of blindness worldwide, making early detection crucial to prevent severe damage to patients. Current studies on cataract detection face limitations, particularly due to the high cost of imaging devices and their limited accessibility for users. In this study, we propose a web application that utilizes a Deep Learning service to analyze fundus images and provide a cataract diagnosis. This application aims to assist healthcare personnel in medical centers lacking specialist ophthalmologists or facing limited resources for cataract diagnosis. We designed the physical architecture of the application using Azure services, enabling its deployment and operation in the cloud. Azure Custom Vision facilitated the training of our model with a dataset of 1446 fundus images, encompassing both cataract and non-cataract cases. Subsequently, we implemented the web application using React.js and Express.js technologies, integrating the Deep Learning model to perform diagnoses through the web interface. The results demonstrated that the model achieved sensitivity, specificity, precision, and accuracy levels exceeding 90%, showcasing that our proposed tool allows for reliable initial cataract diagnoses in patients without the need for high-cost equipment.

Mental health is vital to the development of young adolescents to create strong relationships and resilience, keeping a positive influence on society. The impact of COVID-19 on mental health is anticipated to be significant to the population, especially adolescents by depriving social contact and creating mental disorders. Currently, there are countless Chatbots that give dynamic chatbot services in health. In recent years, neural networks for natural language processing (NLP) have shown to generate more responses learned by the machine. This study aims to implement and compare two Chatbot mobile applications using Neural generative with sentiment models: OpenAI GPT-3 model, and personalized chatbot based on deep Learning, Transformer BERT, and TextBlob model. The dataset for this model was generated from a database of the flow conversation from GPT-3 and the trained bot, frequently asked questions collected at the beginning and end of the term of academic. To analyze, the sentiment it trains the dataset used in the conversation which validates the prediction through a confusion matrix which resulted in test accuracy frequently correct is 70% for Transformer Bert and 68% in TextBlob. To test the usability of the chatbot and application, a survey was conducted on a group of 30 participants, using Chatbot Usability Questionnaire (CUQ) and Linkert scale, it showed that GPT-3 CUQ Mean is 77,71 higher than Deep Learning. As for the Linkert scale, it was verified that 68% of participants perceived that the chatbot was adequate to their concerns, as well that the acceptance rate was 90%.

Ovarian cancer (OC), the most common kind, accounts for more than half of all cases of gynecological cancer in women. The classification of OC might result in several distinct diagnoses (Serous, Mucinous, Endometrioid, Clear Cell). Pathologists use computer-aided diagnosis to assist them make accurate diagnoses. Deep convolutional neural networks (DCNNs) that have previously been trained can recognize, forecast, and classify the different kinds of ovarian cancer. An improved VGG-16 algorithmic structure contains thirteen convolution layers, three of which are linked. In addition, there are five maximum pooling layers and one softmax layer. After capturing 500 images, the model was only recognized with a 50% accuracy rate after training (100 from each class). Using a variety of image processing techniques, we were able to produce a total of 24742 more images from the initial dataset of 500 shots. Only after training on a much bigger dataset did the model’s accuracy increase from 50 to 84%. For the first time, VGG16 histopathological scans are being utilized to diagnose and forecast cancer ovarian tissue.

The Artificial Intelligence (AI) chatbots have emerged as transformative tools across various fields, catalyzing advancements in customer service, healthcare, education, and more. Their versatile nature allows them to engage with users, providing solutions and information while streamlining processes. In customer service, AI chatbots enhance interactions, offering prompt responses and personalized assistance. In healthcare, they facilitate preliminary diagnosis and appointment scheduling, augmenting medical services. Education witnesses their role in interactive learning, delivering knowledge through dynamic conversations. The evolution of AI chatbots in diverse fields showcases their potential to revolutionize how humans and technology interact, intertwining complexity with conciseness, and variability with cohesion, propelling us into an era of enhanced connectivity. This paper provides an insight into what a chatbot is and the types of chatbots. This paper also proposes a classification based on the current trends and uses of chatbots in different fields.

Intelligent assistants are currently categorized as computer programs combined with Artificial Intelligence that interact with the user by voice, messages or images. Thanks to technological advances, intelligent assistants have become widely used in home automation, aiming to automate security, welfare and comfort. This project aims to provide automation for elevators improving the user experience with better security, offering information and user comfort. For this purpose, the application integrates technological tools of facial recognition as part of security and voice activation to command the actuators. Furthermore, the development of the application is oriented to help the multiple users of elevators, especially the group of users with some physical or visual impairment, since it allows the users to mention the floor to which they are going and additionally get a brief overview of what is on that floor, reducing mobility within the elevator, and the search for buttons for the operation of the elevator. The research focused on bibliographic studies on the development of virtual assistants, facial recognition and voice command using free software tools such as Python with its wide range of libraries, and hardware such as PCs, Arduino or Raspberry, depending on the resources needed for the App developed. In the end, a series of tests and satisfactory results with the intelligent virtual assistant are presented to reach the objectives of this study.

It is now easier than ever to “mine” photographs for information and discovers new points of view using techniques. Medical workers now have access to images that can assist them in more swiftly and efficiently diagnosing and treating a broader range of diseases. Dermatologists are working with deep neural networks to discriminate between photographs of healthy skin and those of patients with skin cancer. We have focused our efforts on two important areas of study to gain a better understanding of melanoma. Examining this issue as early in the process as feasible is crucial. Even small changes in dataset properties can have a significant impact on classifier performance. In this part, we’ll discuss the challenges that arise when attempting to adapt what we’ve learned in one environment to another. We believe that repeated training and testing cycles are essential for developing reliable prediction models. Furthermore, a system that is more adaptable and sensitive to changes in training datasets is urgently required. As a result of this, hybrid architecture for service delivery that integrates both clinical and dermoscopic images has been suggested. This approach may be used in a variety of ways, including cloud computing, fog computing, and edge computing. This architecture must be able to analyze large amounts of data while simultaneously speeding up the process of constantly improving its knowledge. In this example, one computer and many distribution mechanisms are employed, demonstrating that the output is obtained in far less time than would be the case with a centralized system. This is in comparison to the guarantees provided by a centralized plan.

Lung cancer mortality is increasing in this country due to a variety of factors, including the country’s increasing industrialization, the buildup of hazardous substances in the environment, and an ageing population. Computed Tomography (CT) scans are routinely used on patients as part of the diagnostic procedure to achieve a conclusive diagnosis of lung cancer. Because of the way X-rays are absorbed by biological tissues, CT scans may detect even the most deeply hidden structures. Pulmonary nodules are lumps or bumps in the lungs that are signs of disease. Because each type of nodule can take many different shapes, the chance of getting cancer varies a lot. In some cases, computer vision models can now help clinicians identify a wide range of medical disorders; in some cases, these models have been found to be more accurate than actual doctors. Deep learning advancements in recent years have made this possible. The disease diagnosis carries a great deal of significance and value for the field due to the numerous opportunities provided by modern technology. This is because the application’s primary function is to serve as a diagnostic tool for a variety of illnesses. Our goals were to improve the accuracy of diagnostic tests and to find diseases earlier. We discovered that the model could detect lung cancer earlier than other approaches already in use. The suggested model includes the following elements: Identifying pulmonary nodules, discussing false positives, and discussing diagnostic uncertainty the number of lung nodules will be reduced by removing “false nodules” and classifying lung nodules as benign or malignant. Throughout human history, new network architectures and loss functions have been developed and implemented. Furthermore, the recommended deep learning mode could be improved, resulting in improved lung nodule identification accuracy. Experiments have shown that the proposed method greatly improves the ratio of accuracy to precision when evaluating the disease being studied.

Since many years ago, scholars have been interested in human-computer communication. The interface between humans as well as computers can be carried out in a variety of ways. Chatbots are a common method for carrying out a conversation. A Chabot system is a computer application that facilitates engaging and simple conversation. Current artificial intelligence techniques perform poorly even when responding to user queries in the most appropriate manner. As a result, industries currently favor rule-based Chabot systems. Artificial intelligence can correctly and consistently anticipate outcomes. The methods have been used in a variety of businesses, academic fields, and contexts. Furthermore, the majority of study has been conducted in industrialized nations, with little work from other economies published. As a consequence, an appropriate research foundation is required for AI applications to be long-term and successful. The goal of this research is to critically evaluate many studies that have used AI in various areas to create a theoretical framework guide for academics and practitioners. This framework will also help to define future research trends in the field. Additional elements of AI that affect the design and functioning of the framework model in practice are the organizational structure and technological specification. A model-based approach that maintains uniformity in research, industry, and academia is required given the present use of AI methodologies. In the present situation, we can use establishing a long-term practice may need a paradigm shift in the framework-based approach. The use case illustrated how the proposed method may be applied for educational purposes to instruct vision-based component orientation recognition using humanizer bot models. Investigating Chatbots’ potential as student mentors.

As evidenced by significant mortality and transmission from the delta and Omicron versions respectively, SARS-COV-2 mutation also causes periodic public worry. We thoroughly examined and condensed several infection disease like COVID-19 treatment, diagnosis, and prevention facets. The biological properties of infectious disease like COVID-19 is first described from the perspective diagnosis. GenBank uses the file extensions .fasta and .gb to store the genomic sequences of several viruses. There were more than 300 unique changes in the genomic nucleotide regions of the four different viral types. Following that, the COVID-19 preceding clinical animal models were reviewed to frame the signs and symptoms of the disease as well as its clinical impacts from patient to patient with the help of therapeutic options and in computational/silicon natural science. Additionally, we looked at the potential and challenges of applying nanomedicine and nanotechnology for identifying, evaluating, and treating infectious disease. This article extensively discusses practically every SARS-CoV-2-associated issue to help readers comprehend the most recent developments. We have used ViroGen algorithm for genomics virus classification for SARS, MERS, Ebola, and SARS-COV-2. We calculate the various performance metrics like precision, accuracy, F1 score, and recall in terms of various model such as SVM, logistic regression, Naïve bayes, etc. with respect to different viruses, and we get 96% accuracy for SARS and 97% accuracy for SARS-COV-2, these suggested technique produces good classification results.

In this paper, we give the focus on the continuous advancement in the domain of Vehicular ad hoc networks (VANET’s) and that is developed as a tool for developing the base for platform intelligent mode in the communication (inter vehicle communications) and needful for enhanced improvement under the performance effect and traffic safety. Here, we have deployed the game theory method in the VANET domain and thereafter categorizes between the hop vehicle and the source vehicle.

The New approach to the VANET analysis we have identified and used the social parameters and hence made it look more novel approach.

“Researching vehicle ad hoc networks poses challenges due to their dynamic nature, the movement of large vehicles, unlimited energy resources, and the evolution of wireless networks. Game theory is commonly employed in wireless networks to explore the interplay between competition and cooperation. In this study, we present a system for ad hoc networks in cars, utilizing game theory to automate car groups and board elections. This approach eliminates the necessity for regular bulk updates. Furthermore, the social behavior of each car is leveraged to create clusters in the car environment. The K-means algorithm in machine learning is applied to develop social cars. The proposed system underwent testing for various characteristics, including CH lifetime, average group lifetime, average number of joins, throughput, and packet loss rate. The results demonstrate that VANET performs exceptionally well, achieving an overall performance ranging from 0.95 to 0.989”.

The purpose of this study is to pinpoint the key variables that influence customers who have never purchased organic food previously in terms of their intention to do so. Attitude, perceived behavioural control, subjective norm, personal norm, & health awareness are the five primary components in this study. Further, this study investigates how machine learning techniques can be used to explore and identify the key determinants influencing the intention to purchase organic food. As the demand for organic products grows, understanding the factors that drive consumers’ decisions to choose organic food becomes increasingly important for both businesses and policymakers. The findings revealed that only four of these five factors—attitude, subjective norm, personal norm, & perceived behavioural control were significant.

This research analyzes how driving style affects the energy consumption of a Kia Soul electric vehicle, studying the signals of the Identification Parameters (PID) in the city of Cuenca, Ecuador. A Real Driving Emissions (RDE) cycle including urban, rural, and highway segments is described, and it is observed that acceleration is a variable directly related to the energy consumption of the electric vehicle. This is more evident in highway areas where speed limits are higher than in urban areas, which makes the vehicle require higher energy consumption. As a result, a 31.14% increase in road consumption can be verified compared to the urban area. The unit density identifies the type of driving (conservative, normal, and aggressive) on the road by means of acceleration profiles and their distribution range. With the implementation of Machine Learning architecture, it is possible to estimate the most important variables, such as accelerator pedal open position (APS), vehicle speed sensor (VSS), and longitudinal acceleration (Ax), in relation to the state of charge (SOC), after applying an ANN to the model. This achieved a prediction with a determination factor of 0.9866 compared to the actual vehicle range.

The growth of electronic data in educational institutions provides an opportunity to extract information that can be used to predict students’ academic performance and dropout rates. This paper provides a survey to explore the current state of research on academic performance prediction using machine learning techniques. A systematic literature search was conducted to identify relevant studies published between 2019 and 2023. The review analyzed studies that used various machine learning algorithms to predict academic performance in different educational contexts. The findings indicate that machine learning models can accurately predict academic performance with a high degree of precision, using a variety of variables such as demographic data, academic history, and student interaction. The review also highlights the challenges of the current research, including the need for collection and preprocessing procedures, and the importance of considering ethical implications related to the use of student data. The findings have important implications for educators, managers, and researchers interested in using machine learning techniques to promote student success.

The academic performance of students is one of the most relevant indicators in the quality of higher education, as well as the importance of the teacher profile in academic activity. This document explores the link between academic performance and the behavior of Engineering students and determines the teacher’s profile based on personality traits. Demographic, academic, and enrollment data of 2001 students were collected, in addition to analyzing the personality traits of four teachers with fuzzy techniques and the Big Five model. The results indicate that academic performance is largely linked to academic factors and specific course activities, and does not depend solely on demographic variables. The teacher’s profile showed a higher percentage in the personality trait of openness to experiences, which can play a crucial role in the academic success of students.

The trade of horticultural products is a crucial sector in the local economy of Lima, Peru. Microenterprises dedicated to this activity face various challenges, including demand volatility. This volatility can decrease the likelihood of generating profits and impact the stability of the business, primarily due to the challenges associated with adjusting selling prices. To address this issue, our proposal is based on implementing the XGBoost algorithm, which has the capability to handle heterogeneous data and variables of different types. This algorithm leverages historical data to provide accurate and up-to-date price recommendations for horticultural products. This, in turn, enables micro-entrepreneurs to make informed decisions when setting prices, thereby achieving expected benefits and enhancing their competitiveness. The integration of our project with microenterprises in Lima has the potential to mitigate the risk of economic losses by offering greater accuracy in predicting future market prices. Through the development of our project, we have achieved a high level of accuracy in forecasting future prices, reaching a minimum of 90% when compared to actual prices.

The constant increase in vehicles and their consequent production of pollutant emissions from mobile sources is a global predicament. In response, efforts are underway to mitigate fuel consumption. This metric is influenced not only by the vehicle type and fuel choice but also by strategic decisions such as maintenance, route selection, and driver behavior. Driving style is a parameter that is often ignored but has great repercussions on energy costs. Moreover, Ecuador lacks studies on driver behavior in real conditions. This study conducted dynamic tests to gather data through a data logger device. Subsequently, the collected data underwent a preprocessing stage to extract relevant features. These predictors were then employed to train a decision tree model for discerning between normal and aggressive driving styles. Our findings reveal that during urban driving, the aggressive style results in an average consumption of approximately 15% higher than the normal style. In rural settings, this difference is around 13%, with aggressive driving consuming 0.644 l/h more. However, during highway driving, no significant difference in average fuel consumption was observed between the two driving styles.

This paper presents the development of a conversational agent embedded in a mobile application aimed at improving the safety of medication use in patients with diabetes, a highly prevalent chronic disease. One of the main challenges in the management of diabetes is the prevention of drug interactions and the understanding of how food can affect treatment.

The conversational agent uses a database of drug, food, and pharmacological interaction information to provide both preventive and informative recommendations. In the first scenario, it alerts users to possible interactions between the medications they are planning to take. In the second scenario, it provides information on contraindications, suggests food alternatives, and points out possible drug-food and drug-drug interactions. The design of the conversational agent allows it to understand and answer questions in natural language, making it easier for users to intuitively access relevant information. The usability of the tool was evaluated, resulting in an overall positive perception by users. In addition, the system identifies drug-drug and drug-food interactions and provides useful recommendations as indicated by experts.

Although areas for improvement were identified and specific adaptations are needed to fully meet the needs of users, the implementation of the agent through a mobile application has the potential to prevent dangerous interactions for patients with diabetes and significantly improve their quality of life.