Muhammad Ikhsan Mokoagow, Arief Purnama Muharram, Ahmad Zufar Manthovani, Marina Epriliawati
Critical illness is a condition that requires special attention, as it often involves life-threatening situations with a higher mortality rate. Attention is further focused on diabetic patients, who have recently become common comorbid cases among critically ill patients. Patients with these conditions are often prone to hyperglycemia and hypoglycemia due to the nature of critical illness and diabetes itself. Intensive Insulin Therapy (IIT) is known to be beneficial in maintaining blood glucose levels at the optimal target in such patients. However, the application of IIT is not straightforward. To date, there is no standardized IIT protocol, and its use is highly dependent on the expertise of each clinician. Moreover, each protocol does not always guarantee the maintenance of blood glucose levels because it depends on the patient's response to insulin, while the protocol is designed to be general. With the rapid advancement of Artificial Intelligence (AI), particularly Machine Learning (ML), a personalized IIT can be developed to address this issue. With the power of data, AI algorithms can analyze and assist in creating personalized IITs according to the patient's condition, with the goal of maintaining 2 their blood glucose levels. In this review, we present a future directive on the potential role of AI in creating personalized IITs for critically ill diabetic patients, as well as the challenges to be overcome in the development process.
Keywords: Artificial Intelligence, Machine Learning, Intensive Insulin Therapy, Diabetes Mellitus, Critically Ill Patients
Arief Purnama Muharram, Ayu Purwarianti
Submitted to Journal of ICT Research and Applications Spotlight Master's thesis
Automated fact-checking is a key strategy to overcome the spread of COVID-19 misinformation on the internet. These systems typically leverage deep learning approaches through Natural Language Inference (NLI) to verify the truthfulness of information based on supporting evidence. However, one challenge that arises in deep learning is performance stagnation due to a lack of knowledge during training. This study proposes using a Knowledge Graph (KG) as external knowledge to enhance NLI performance for automated COVID-19 fact-checking in the Indonesian language. The proposed model architecture comprises three modules: a fact module, an NLI module, and a classifier module. The fact module processes information from the KG, while the NLI module handles semantic relationships between the given premise and hypothesis. The representation vectors from both modules are concatenated and fed into the classifier module to produce the final result. The model was trained using the generated Indonesian COVID-19 fact-checking dataset and the COVID-19 KG Bahasa Indonesia. Our study demonstrates that incorporating KGs can significantly improve NLI performance in fact-checking, achieving the best accuracy of 0,8616. This suggests that KGs are a valuable component for enhancing NLI performance in automated fact-checking.
Keywords: Fact-Checking, Deep Learning, Natural Language Inference, Knowledge Graph, COVID-19
Arief Purnama Muharram, Farhan Hilmi Taufikulhakim, Ayu Purwarianti
2023 IEEE International Biomedical Instrumentation and Technology Conference (IBITeC). doi: 10.1109/IBITeC59006.2023.10390908
COVID-19 is an acute respiratory disease that has become a pandemic worldwide. Many studies have been conducted to enhance our understanding of COVID-19. However, the abundance of information obtained from these studies has resulted in information overload. In this study, we purposed a simple COVID-19 Knowledge Graph in Bahasa Indonesia as a way to reconstruct knowledge to combat this information overload. We used Bahasa Indonesia in our study to explore its potential for constructing a Knowledge Graph (KG). The construction of our KG involved manual curation of medical literatures and annotation of entities and relationships by the domain experts. The KG was implemented using Neo4J version 5. We successfully demonstrated our COVID-19 KG, which consists of 240 nodes and 276 relationships with 15 and 14 node and relationship labels respectively. Accessing the information within the KG is made effortless through the use of Cypher queries in Neo4J. Further research is still needed to develop the KG into a larger and better one. However, our COVID-19 KG can serve as a basis for further development.
Keywords: COVID-19, Knowledge Graph, Bahasa Indonesia
Arief Purnama Muharram, Fahmi Sajid
2023 International Conference on Electrical Engineering and Informatics (ICEEI). doi: 10.1109/ICEEI59426.2023.10346215
Diabetes mellitus, particularly type-2 diabetes, remains a prevalent health issue, raising concerns due to its associated risk of complications. Among these, cardiovascular complications pose a significant threat, exhibiting high morbidity and mortality rates. Health screening plays a pivotal role in stratifying the risk levels of diabetes patients, facilitating proactive measures to prevent the progression of complications. As such, the primary objective of this study is to develop a predictive model system for assessing cardiovascular risk in diabetes patients. Our study used the Cardiovascular Disease dataset and conducts experiments with various supervised machine learning algorithms, such as Naive Bayes, decision tree, random forest, AdaBoost, and XGBoost. The results reveal that ensemble learning algorithms based on boosting, particularly AdaBoost and XGBoost, outperform other supervised machine learning methods. However, even with the best performance achieved using the dataset, the accuracy stands at 0.71, and the F-1 score is 0.69, which is still acceptable for screening purposes. Although these results provide valuable insights, indicating individuals at higher risk for cardiovascular complications in diabetes, further improvements are needed to enhance early prevention strategies.
Keywords: Diabetes Mellitus, Cardiovascular Complications, Machine Learning, Naive Bayes, Decision Tree, Random Forest, AdaBoost, XGBoost
Arief Purnama Muharram, Dicky Levenus Tahapary, Yeni Dwi Lestari, Randy Sarayar, Valerie Josephine Dirjayanto
2023 10th International Conference on Advanced Informatics: Concept, Theory and Application (ICAICTA). doi: 10.1109/ICAICTA59291.2023.10390334
Diabetes, especially T2DM, continues to be a significant health problem. One of the major concerns associated with diabetes is the development of its complications. Diabetic nephropathy, one of the chronic complication of diabetes, adversely affects the kidneys, leading to kidney damage. Diagnosing diabetic nephropathy involves considering various criteria, one of which is the presence of a pathologically significant quantity of albumin in urine, known as albuminuria. Thus, early prediction of albuminuria in diabetic patients holds the potential for timely preventive measures. This study aimed to develop a supervised learning model to predict the risk of developing albuminuria in T2DM patients. The selected supervised learning algorithms included Naive Bayes, Support Vector Machine (SVM), decision tree, random forest, AdaBoost, XGBoost, and Multi-Layer Perceptron (MLP). Our private dataset, comprising 184 entries of diabetes complications risk factors, was used to train the algorithms. It consisted of 10 attributes as features and 1 attribute as the target (albuminuria). Upon conducting the experiments, the MLP demonstrated superior performance compared to the other algorithms. It achieved accuracy and f1-score values as high as 0.74 and 0.75, respectively, making it suitable for screening purposes in predicting albuminuria in T2DM. Nonetheless, further studies are warranted to enhance the model's performance.
Keywords: Diabetes, Albuminuria, Supervised Learning, Machine Learning, Deep Learning
Anggi Septia Irawan, Bence Márton Döbrössy, Mengesha Srahbzu Biresaw, Arief Purnama Muharram, Dávid Szilárd Kovács, Edmond Girasek
[Background] Electronic health (E-Health) and mobile health (M-Health) have emerged as promising solutions to address the scarcity of healthcare workers in developing countries. This systematic literature review aims to comprehensively explore the utilization of E-Health and M-Health tools in pediatric care among parents and healthcare workers in these regions.
[Methods]
A literature search was conducted on the PubMed and Scopus databases in January 2023, covering the period from 2013 to 2022. The search adhered to the PRISMA guidelines. The PECOS format outlined by PROSPERO was used to determine the eligibility of systematic reviews and primary studies, with no restrictions on study designs or methodologies. Eligible articles comprised original research published in peer-reviewed open-access journals, limited to the English language. The methodological quality of the included articles was independently assessed by authors using the CASP checklists to evaluate reporting quality.
[Result]
The initial search yielded 334 publications, of which 16 met the inclusion and exclusion criteria for the review. These comprised 12 Random Control trials and 4 Qualitative-Quantitative studies. The CASP criteria were applied to all studies, and those with a moderate to high level of methodological quality were included and reported. The reviewed publications described various types of E-Health and M-Health tools, with a majority (50%) of the studies conducted in Asia. The target users in the publications were diverse, with 37% focusing on mothers, 25% targeting pregnant women, and 19% targeting healthcare workers.
[Conclusions]
The review highlights a diverse range of tools, including mobile applications, websites, SMS, and phone calls, with a particular focus on breastfeeding, vaccination, and child growth. The findings emphasize the importance of healthcare worker participation and the trust placed in information from relatives to influence the effectiveness of these tools. Moreover, the study underscores the need for intimate discussions when addressing sensitive topics like HIV and contraceptives. This review enhances our understanding of the role of E-Health and M-Health tools in pediatric care in developing countries. It highlights their potential to bridge healthcare gaps and promote wider access to quality care, addressing the challenges posed by limited healthcare resources in these regions.
Keywords: E-Health, M-Health, Digital Health, Pediatric Care, Developing Countries
Arief Purnama Muharram, Hollyana Puteri Haryono, Abassi Haji Juma, Ira Puspasari, Nugraha Priya Utama
2023 IEEE International Conference on Data and Software Engineering (ICoDSE). doi: 10.1109/ICoDSE59534.2023.10291842 Spotlight Insight
Reading and interpreting chest X-ray images is one of the most radiologist's routines. However, it still can be challenging, even for the most experienced ones. Therefore, we proposed a multi-model deep learning-based automated chest X-ray report generator system designed to assist radiologists in their work. The basic idea of the proposed system is by utilizing multi binary-classification models for detecting multi abnormalities, with each model responsible for detecting one abnormality, in a single image. In this study, we limited the radiology abnormalities detection to only cardiomegaly, lung effusion, and consolidation. The system generates a radiology report by performing the following three steps: image pre-processing, utilizing deep learning models to detect abnormalities, and producing a report. The aim of the image pre-processing step is to standardize the input by scaling it to 128x128 pixels and slicing it into three segments, which covers the upper, lower, and middle parts of the lung. After pre-processing, each corresponding model classifies the image, resulting in a 0 (zero) for no abnormality detected and a 1 (one) for the presence of an abnormality. The prediction outputs of each model are then concatenated to form a 'result code'. The 'result code' is used to construct a report by selecting the appropriate pre-determined sentence for each detected abnormality in the report generation step. The proposed system is expected to reduce the workload of radiologists and increase the accuracy of chest X-ray diagnosis.
Keywords: Chest X-Ray, Radiology, Medical Report, Multimodel, Deep Learning
Anindya Pradipta Susanto, Hariyono Winarto, Alessa Fahira, Harits Abdurrohman, Arief Purnama Muharram, Ucca Ratulangi Widitha, Gilang Edi Warman Efirianti, Yehezkiel Alexander Eduard George, Kevin Tjoa
Informatics in Medicine Unlocked. doi: 10.1016/j.imu.2022.101017
Emerging technologies powered by artificial intelligence (AI) have sparked hope of achieving better clinical outcomes among patients. One of the trends is the use of medical image recognition systems to screen, diagnose, or stratify risks of diseases. This technology may enhance sensitivity and specificity and thus, improve the accuracy and efficiency of disease diagnosis. Therefore, it is important and beneficial for healthcare providers to understand the basic concepts of AI so that they can develop and provide their own AI-powered technology. The purpose of this literature review is to provide (1) a simplified introduction to AI, (2) a brief review of studies on medical image recognition systems powered by AI, and (3) discuss some challenging aspects in this field. While there are various AI-powered medical image recognition systems, this paper mainly discusses those integrated in smartphone apps. Medical fields that have implemented image recognition models in smartphones include dermatology, ophthalmology, nutrition, neurology, respiratology, hematology, gynecology, and dentistry. Albeit promising, AI technology may raise challenges from the technical and social aspects of its application. Notable technical issues are limited dataset access and small datasets, especially for rare diseases. In a social context, the perspectives of all involved parties (physicians, patients, and engineers) must be considered.
Keywords: Artificial Intelligence, Deep Learning, Medical Image Recognition, Smartphone Applications
Arief Purnama Muharram, Anindya Prameswari Ekaputri, William Fu, Hollyana Puteri Haryono, Adriel Gustino Parlinggoman Situmorang
Harvard Dataverse. doi: 10.7910/DVN/LNWKPY
Keywords: Graves' Disease, Chatbot, Natural Language Processing
Arief Purnama Muharram, Harits Abdurrahman, Ucca Ratulangi Widitha, Alessa Fahira, Muhammad Gilang Edi, Anindya Pradipta Susanto, Hariyono Winarto
The International Congress of Asia Oceania Research Organization on Genital Infections and Neoplasia 2021
[Background]
Cervical cancer is one of the most common types of cancer among women, caused by thehuman papillomavirus (HPV) infection, commonly types 16 and 18 (high-risk types). Globally, there were approximately 570,000 cases and 311,000 deaths in 2018, with the incidence ranging around 13.1 per 100,000 women and varied widely among countries. In Indonesia, there were 36,633 new cases and 21,003 deaths reported in 2020, making it the second most common malignancy in women after breast cancer. Routine screening, especially in high-risk populations, is a key to prevent disease progression by prompt treatment at an earlier stage. The visual inspection with acetic acid (VIA) is considered costeffective and widely available as a practical test for the national cervical cancer screening program. The test is also considered to be safe and can be done by trained health workers. However, interpretation of the result requires skills and is highly subjective. Thus, the result’s interpretations may be slightly different among workers. Emerging artificial intelligence technology may provide the solution to minimize the variability of those results. However, designing such technology requires big, validated data. Therefore, this research aims to build a web-based medical image repository system to make it easier to collect and manage VIA-test images dataset for creating and training such robust artificial intelligence technology. The system’s key feature lies in the automatic voted image categorization system, of which there were three categories (positive, negative, or inconclusive) based on the ‘vote’ (label) judgement given to the image.
[Methods]
The system was built on Laravel PHP framework version 7 and deployed on Intel® Xeon® CPU E5-2640 @2.40 GHz, 6 gigabytes of RAM server running Ubuntu 20.04 LTS, PHP 7.4.3, MySQL 8.0.23, and NginX 1.18.0. There were 868 collected unlabeled VIA-test images from our database stored on the system. We involved ten medical consultants from the Department of Obstetrics and Gynecology, Dr. Cipto Mangunkusumo National General Hospital, Faculty of Medicine, Universitas Indonesia, Jakarta. Those images were further divided into five randomized groups, with each group consisting of around 173 images. Each group of images would be interpreted and graded by two randomized consultants based on the VIA-test result, whether it is acetowhite- positive or negative, through the system. If the result was inconclusive, defined by the previous consultants' different interpretations (positive-negative or negative-positive), a third independent consultant would be involved in grading the image. The gradings were conducted from August to December of 2020.
[Results]
The system was named the “Simple Acetowhite Labeling” website, and the source code was available open for access through CerviCam Research Group GitHub Repository (https://github.com/CerviCam/Simple-Acetowhite-Labeling-Web-Application). After image grading process by the consultants, 420 images were categorized as positive VIA-test, while the 448 rest were categorized as negative VIA-test. The categorized images could be archived together and downloaded as a dataset (positive or negative VIA- test dataset) through the system menu. The dataset was then further used to train artificial intelligence models for interpreting results of VIA-test images.
[Discussion]
The “Simple Acetowhite Labeling” system provides practicality in accessing and categorizing VIA-test image collection. Being a website-based platform, the system is accessible through internet from any physical location. The system also provides ease of adaptability for medical image repository for other health institutions.Other than the aforementioned advantages, one of the main challenges during the system development was in designing a clear and intuitive user interface that was easily understood by the consultants as the users. Creating a user journey guideline and planning an intensive training session for the consultants are deemed necessary to overcome this challenge. Moreover, the system is also dependent on expert opinions which is subjective by nature. Consequently, objective data validation could not be guaranteed. Further research should be conducted to develop a more objective data validation mechanism.
[Conclusions]
The development of the “Simple Acetowhite Labeling” system as a web-based medical image repository system is an initial step toward creating artificial intelligence technology in automatizing VIA-test result's interpretation. The system was designed to manage VIAtest image collection and further generate datasets that could be used to create and train artificial intelligence models.
Keywords: Cervix Uteri, Uterine Cervical Neoplasms, Visual Inspection with Acetic Acid, Medical Informatics Application
Arief Purnama Muharram, Dicky Levenus Tahapary, Pradana Soewondo
International Diabetes Federation (IDF) Congress 2019
The prevalence of DM in Indonesia, the fourth most populous country in the world, has increased by more than 50% in the past 5 years, from 6,9% in 2013 to 10,9% in 2018, of which most of the were Type-2 (T2DM). Furthermore, >70% of cases were undiagnosed and among those receiving treatment, only <30% achieved the blood glucose target. These factors lead high rates of complications and health cost which further complicates the problem of DM in Indonesia. Diabetes Self-Management Education (DSME) is one of strategy to overcome this problem and well-known to give good positive outcome. In past few years, many smartphone applications are developed to support DSME-behavior among DM patients. However, none of these applications/approach are specifically designed for DM patients in Indonesia. The aim of this research is to develop and validate DM EduCorner as a standardized smartphone-based application for facilitating diabetes education among DM patients in Indonesia.
Keywords: Diabetes Mellitus, Diabetes Education, Smartphone Application
Arief Purnama Muharram, Pradana Soewondo, Dicky Levenus Tahapary
20th ASEAN Federation of Endocrine Societies (AFES) Congress 2019 Spotlight Bachelor's thesis
[Introduction]
The rapid development of smartphone technology nowadays has enabled a new way of diabetes self-empowerment through the smartphone application usage. This study is aimed to obtain an overview of how smartphone and smartphone applications are used for diabetes management among Type-2 Diabetes Mellitus (T2DM) patients in RSUPN Dr. Cipto Mangunkusumo Jakarta (RSCM), a tertiary care and a national referral hospital in Indonesia.
[Methodology]This cross-sectional study was conducted in the Integrated Diabetic Clinic RSCM during the 2nd-to-3rd week of May 2019 by using a short questionnaire, of which assessed the level of smartphone ownership and smartphone application usage for diabetes management.
[Results]Thirty-one respondents participated in this study. The average age was 59 years-old and most of them were either retired (13/31, 41.9%) or not working (13/31, 41.9%). Only 11 respondents had a higher degree of education. While most of the respondents (18/31, 58.1%) had a basic monthly income <1 million IDR (60 USD), majority of respondents (27/31, 87.1%) had a smartphone, of which all of them were using Android. Only one respondent used it for diabetes management, while most of them used it only for standard communication purpose. This was due to the lack of information on available diabetes application.
[Conclusion]The use of smartphone among T2DM patients in our tertiary care hospital was high despite their low socioeconomic status. However, the smartphone application usage for diabetes management was very low, necessitating the need of information dissemination related to the potential benefit of diabetes application to all T2DM patients.
Keywords: Smartphone, Diabetes Application, Type 2 Diabetes Mellitus, Self-Management
Harits Abdurrohman, Robih Dini, Arief Purnama Muharram
Seminar Nasional Instrumentasi, Kontrol dan Otomasi (SNIKO) 2018
The HAM10000 Dataset merupakan koleksi besar citra dermatoskopi untuk lesi kulit berpigmen yang umum. The HAM10000 Dataset terdiri atas 10.015 data citra lesi kulit berpigmen yang terbagi untuk penyakit Bowen, karsinoma sel basal, benign keratosis-like lesion, dermatofibroma, melanoma, melanocytic nevi, dan lesi vaskular. Data citra yang terdapat dalam dataset telah terkonfirmasi baik melalui histopatologi, pemeriksaan follow-up, konsensus pakar, maupun konfirmasi melalui in-vivo confocal microscopy. Pada penelitian ini kami melakukan pengujian performa terhadap model deep learning dan melakukan evaluasi. Tahap pre-processing citra meliputi analisis distribusi citra pada setiap kelas lesi, pengelompokan ulang kelas lesi berdasarkan letak pada bagian tubuh, dan augmentasi citra. Oleh karena keterbatasan data citra setelah dilakukan analisis distribusi maka model yang dibangun pada penelitian ini hanya berfokus pada kelas lesi untuk abdomen, punggung, ekstremitas atas dan bawah. Evaluasi ini dilakukan terhadap beberapa metode yang terkenal InceptionV3, MobileNet dan MobileNetV2. Ukuran performa yang dilakukan meliputi analisis confusion matrix yakni dengan mengambil nilai precision dan recall, dan f1-score.
Kata kunci: Deep Learning, Performance, The HAM10000 Dataset, Skin Lesion
Arief Purnama Muharram, Rachma Hadiyanti, Rahmadian Tio Pratama, Dicky Levenus Tahapary
14th Jakarta Endocrine Meeting 2018 Student Creativity Project
Diabetes Melitus (DM) merupakan sekelompok penyakit gangguan metabolik dengan karakteristik hiperglikemia kronis yang terjadi baik akibat kelainan pada sekresi insulin, kerja insulin, atau keduanya. Dalam penatalaksanaannya, penatalaksanaan DM tipe 2 tidak hanya terfokus pada terapi farmakologis saja, tetapi harus mencakup 4 pilar utama pengelolaan DM yang meliputi edukasi, diet, aktivitas fisik dan olahraga, dan terapi farmakologis (pengobatan). DM NutriAssist (Diabetes Mellitus Personal Nutrition Assistant) merupakan aplikasi asisten pribadi berbasis Android yang dikembangkan dengan tujuan untuk membantu manajemen nutrisi dan kontrol metabolik pasien DM tipe 2. Aplikasi ini dilengkapi dengan fitur-fitur yang dapat membantu manajemen nutrisi dan kontrol metabolik pada pasien DM tipe 2 yang meliputi, 1) fitur pencatatan dan analisis gula darah, kolesterol, dan tekanan darah; 2) fitur pencatatan dan analisis nutrisi makanan yang dikonsumsi oleh pengguna; 3) fitur pencatat dan penjadwalan kunjungan rutin ke dokter; dan 4) fitur pojok edukasi yang berisi tentang artikel seputar penyakit DM beserta pengelolaannya. Aplikasi ini didesain untuk dapat dijalankan pada smartphone Android dengan versi minimal 7.1 (Nougat), RAM 1 GB, dan kapasitas penyimpanan tersisa minimal 100 MB. Saat ini, aplikasi ini masih dalam tahap evaluasi dan pengembangan lebih lanjut, khususnya dalam hal optimasi fitur dan pengembangan basis data nutrisi makanan. Akan tetapi, diharapkan kedepannya aplikasi ini dapat menjadi solusi praktis untuk membantu manajemen nutrisi dan kontrol metabolik pasien DM, khususnya DM tipe 2, di Indonesia. Kata kunci: Aplikasi Android, diabetes melitus, kontrol metabolik, manajemen nutrisi