International Workshop on Multidisciplinary Research in Artificial Intelligence 2018

November 12, 2018 

Kathmandu, Nepal

Temporal logic-based fuzzy decision support system for diagnosis of Rheumatic Fever and Rheumatic Heart Disease 

Prof. Prakash Raj Regmi & Dr. Sanjib Raj Pandey

Joint Presentation


Abstract :

This is a collaborative project between the Nepal Heart Foundation (NHF) and the University of Greenwich (UoG), United Kingdom (UK). Our mutual aim, agreed at the outset, has been to analyse, design and develop a cost effective Clinical Decision Support System (CDSS) for diagnosis and recognition of Acute Rheumatic Fever (ARF) and Rheumatic Heart Disease (RHD) at an early stage by developing/adopting UK’s and NHF’s treatment practices and procedures that would be appropriate for the Nepalese environment and lifestyle. The Application we developed was designed for use by community health workers and doctors in the rural areas of Nepal where laboratory facilities, expert services and technology are often deficient. 


The research undertaken investigated three problems that previously had not been addressed in the Artificial Intelligence (AI) community. These are: 1) ARF in Nepal has created a lot of confusion in diagnosis and treatment, due to the lack of standard procedures; 2) the adoption of foreign guidelines is often not effective and does not suit the Nepali environment and lifestyle and 3) the value of combining (our proposed) Hybrid Approach (Knowledge-based System (KBS), Temporal Theory (TT) and Fuzzy Logic (FL)) to design and develop an application to diagnose ARF cases at an early stage in English and Nepali. The overall ARF diagnostics performance and accuracy was 99.36%. Therefore, the experiments and evaluations of our ARF Diagnosis Application proved that it was logically and technically feasible to employ the Hybrid Approach for developing a new and practical ARF Diagnosis Application.

Sanjib Raj Pandey

Prabin Raj Shakya

Kathmandu University Dhulikhel Hospital


Abstract :

Adherence to hand hygiene is the most effective strategies to prevent hospital acquired infection . Non adherence to hand hygiene can lead to prolonged hospital stay, economic burden and mortality of patient. Hand hygiene reduces Healthcare Associated Infections (HAIs) and related economical, psychological and physical burden of patients. However, knowing all this facts, Health Care Workers (HCWs) are still non compliant to hand hygiene. Studies have shown that interventions such as performance feedback, educational intervention with the involvement of leaders are effective ways to improve hand hygiene compliance of HCWs.

The aim of this study is to implement a dashboard on results of hand hygiene compliance audit as visual performance feedback to HCWs and measure its effectiveness.



An interventional study will be conducted on the compliance of hand hygiene (either using disinfectants or soap and water) for the five moments of hand hygiene given by World Health Organization (WHO). Health care workers working in intensive care units will be observed during their routine care in day shift. Technique of hand hygiene will also be measured through hand washing or hand hygiene with alcohol based disinfectants. Dashboard on audit will be prepared and will be displayed in individual units for intervention group on a weekly basis.

Data are electronically collected using ODK collect and kobotoolbox. The dashboard was created in R Shiny App.

Data Driven Behaviors Change: Step-wedge RCT on Hand Hygiene Compliance

D. Parineeta Upadhyaya

The British College, Nepal

Being the head of the National information Technology Center, She is responsible for performing a number of tasks. They are interacting with various government organizations at regular intervals for understanding the scope of the projects and identifying their needs, presentations of the solutions in a written format, monitoring and designing new software, arranging training sessions for other consultants, etc. besides regular activities like administrative and technical works 

Application of AI in Transforming health, education, agriculture, and economy of Nepal

Swarup Rai Dahuguna

Swarup Raj Dhungana

The British College, Nepal

The main aim of the research is to identify and evaluate the object detection models trained for MSCOCO data sets provided by the TensorFlow framework. TensorFlow is one of the newest open source machine learning library provided by Google which was initially released on November, 2015. The TensorFlow Object Detection API is a framework built on top of TensorFlow that makes the process of constructing, training and deploying of the object detection model very smooth. MSCOCO data sets are the training data sets which includes 80 different categories of objects which are further used for training to finally develop models. There are varieties of models trained on the coco data sets; ssd_inception_v2_coco, faster_rcnn_resnet101_coco, rfcn_resnet101_coco and so on. 


In the research, at first we are implementing one of these available object detection models to detect several objects in one particular image and later detecting objects in the same image using other different model and finally evaluate and compare the results of their performance. The evaluation would basically base on three different pillars; the number of objects being identified, the accuracy of the object being identified and the time taken to load the models. JAVA with Object Oriented approach is mainly used for the research along with a bit of Python in Linux OS.

Comparison of Object Detection Models in TensorFlow

© 2018 - 2023 by CMRAI.

CMRAI - Incorporated on 10 August 2017,    Company No.10909300

Designed by Primafelicitas Ltd.


Big Data /Deep Learning & AI : sam@c-mrai.org

AI and Health Informatics : sanjib@c-mrai.org

AI and Security : john@c-mrai.org

  • Black Facebook Icon
  • Black Twitter Icon
  • Black Instagram Icon


Unit 9 44 Plumstead High Street

London, SE18 1SL

Email : info@c-mrai.org