Best Artificial Intelligence Solution in Healthcare

Honors the best artificial intelligence solution that excels in analysis, interpretation, and comprehension of complicated medical and healthcare data to improve patient outcomes. The winning solution has the ability to gain information, process it and give a well-defined output to the end-use through machine and deep learning.

EyeOn, EyeTech Digital Systems - EyeTech Digital Systems

EyeTech Digital Systems' EyeOn platform combines next-gen eye-tracking technology with the power of a portable, lightweight tablet, making it the fastest, most accurate device for Augmentative and Alternative Communication (AAC). Designed to give a voice to non-verbal patients with conditions such as cerebral palsy, autism, ALS, muscular dystrophy, stroke, traumatic brain injuries, spinal cord injuries and Rett syndrome, EyeOn empowers users to communicate and improve their quality of life, hands-free with the power of their eyes.

  • Go To EyeTech Digital Systems' Web Site
  • Go To Product Site
  • Supporting Video
  • DrAid, VinBrain LLC - VinBrain LLC

    VinBrain LLC is a medical AI-focused company established in April 2019. In June 2020, we launched DrAidTM, our flagship AI-powered healthcare product.

    DrAidTM provides AI-based automated chest X-ray diagnosis and screening to radiologists and contains comprehensive features to optimize their daily tasks.

    DrAid’s AI is trained on our self-curated dataset of 1.3 million X-ray images sourced from public and private sources. It applies our patented labelling methodology and utilizes clean medical image data process supported by a group of top expert radiologists. We have also utilized our proprietary graph convolution neural network (G-CNN) deep learning and active learning technologies. Each time a diagnosis is completed, DrAidTM combines knowledge from the image analysis and medical report to further train its model through a feedback loop system.

    Currently, DrAidTM can detect the 20 most common chest X-ray abnormalities and diseases including COVID-19, tuberculosis, pneumonia, and chest fracture with an average F1 score of 88%. In our A/B tests, we observe that radiologists achieve between 9-25% higher accuracy by diagnosing with our product.

    In hospitals, DrAidTM can easily integrate with their systems, e.g. picture archiving and communication systems (PACS or medical imaging machines, to transfer images to the cloud and process them in near real-time. DrAidTM can also integrate directly with health information systems (HIS) to minimize onboarding efforts and costs. Front-end, we support web-based, desktop-based, Android and iOS-based apps, as well as via an DrAid™ Cognitive API service for any 3rd party integration.

    Once integrated, image upload and interpretation take place in the VinBrain cloud. To ensure privacy protection for patients and hospitals, we have built-in AI pre-processor models to remove any PII (Personal Identifiable Information) to stay compliant with medical privacy laws. As for security, data in DrAidTM is encrypted in transit via HTTPS and at rest in the cloud.

    In detailed, information from Xray machines is transferred to VB Loader/VB OneBox following DICOM standards (information is still in the hospital network system). VB Loader/VB OneBox will remove any PII on DICOM file (including blackening patient information on the images and deleting patient information from DICOM medata) and use HTTPS to send DICOM to DrAid Cloud. Doctors will access DrAidTM by AI conclusion browser from Chrome via HTTPS for diagnosis.

    Our data center on PACS cloud has acquired the following security certificates: CIS Benchmarks, ISO 27001:2013: Information technology — Security techniques — Information security management systems — Requirements, ISO 22301:2019: Security and resilience — Business continuity management systems — Requirements, ISO 27017:2015: Information technology — Security techniques — Code of practice for information security controls based on ISO/IEC 27002 for cloud services; and PCI DSS. DICOM server follows two standards which are DICOM and HL7.

    In a situation where the hospital does not have existing PACS, VinBrain will store data in at least two independent places (offline and online) within Vietnam, and back up every 5 minutes with changes occurring. Data is encrypted using 256-bit AES encryption.

    After analyzing the uploaded X-ray, DrAidTM sends notifications to doctors if it detects any abnormalities in the images. Every finding from DrAidTM comes with a probability score, a standardized description and a heatmap identifying abnormal areas. For patients with multiple records, DrAidTM can compare and give detailed information about size and location changes over time.

    Doctors can also further interact with and analyze the medical image using various image analysis tools. Once the diagnosis is finalized, DrAidTM automatically generates a medical report that can be modified by typed or voice-based commands. Our medical report follows international standard format using NLP techniques.  The accuracy score of our voice recognition model in Vietnamese language medical domain is 6.6 WER (Word Error Rate) which is much better than Google’s general purpose Vietnamese voice recognition model at 22.1 WER in medical domain. Doctors can also share the images and report with their patient or refer the case to a second doctor by a secured link and QR code.

    Currently DrAid has been deployed in 82 hospitals and being actively used by 464 doctors in Vietnam. Till now more than 325,000 X-rays have been diagnosed with DrAid’s help. We are on track for deployment in 150 hospitals in Vietnam and 7 hospitals in Myanmar in 2021. We have also started US FDA submission process and hope to deploy DrAid to US hospitals/clinics after FDA approval.

    We are also organizing a joint X-ray diagnosis competition with Stanford for improving state of the art AI techniques in the world. An international organization called Friends for International TB Relief has selected us among top partners in the world for automated AI diagnosis of TB. It demonstrated diagnosis for TB with remarkable specificity (99%), sensitivity (90.1%) and F1 Score (90.5%). We are also currently working with FIT and the division of global TB of CDC USA on the computer-aided diagnosis protocol using AI in TB triage testing. The protocol is expected to be approved by CDC Atlanta in the 2nd Quarter of 2021. Once approved, it will be submitted to WHO and be officially recommended as a diagnosis tool for TB.

    In recognition of these achievements, we recently won the “Best Digital Solutions Make-in-Vietnam” award from the Government of Vietnam.

  • Go To VinBrain LLC's Web Site
  • Go To Product Site
  • Supporting Video
  • Supporting PDF
  • VinBrain LLC Logo