Tel Aviv Sourasky Medical Center has been a pioneer in the realm of digital health information for the last decade. We have harnessed a wealth of electronic medical records, monitoring data, imaging data, laboratory results, as well as operational, logistics, and financial information to create our "data ocean". This robust and diverse dataset of structured and unstructured data has enabled us to stay at the forefront of advancements in healthcare and drive innovation in the field of digital health.
Available Data Types
Have access to the vast data ocean of Tel Aviv Sourasky Medical Center digitally collected since 2007:
Electronic health records (EHRs( comprise both structured and unstructured data.
The easily accessible structured data includes patient demographics, vital signs, diagnoses, medications, lab test results, habits, procedures, sensitivities, and more.
Additionally, EHRs consist of free text that can be anonymized upon request.
Unstructured data contains recordings (EEG, ECG) and Medical imaging data such as X-rays, computed tomography (CT) scans, magnetic resonance imaging (MRI), ultrasound etc. These images provide visual representations of anatomical structures, allowing for the identification and assessment of abnormalities, injuries, and diseases. Radiology reports, generated by radiologists, accompany imaging data, offering expert interpretations and findings.
Pathology data includes tissue samples, biopsies, and cytology slides. Pathology reports document the findings, including histopathological descriptions, immunohistochemistry results, and molecular testing outcomes.
Collectively, these diverse types of medical data provide a comprehensive view of a patient's health status, enabling healthcare professionals, researchers, and data scientists to derive insights, develop predictive models, and improve healthcare delivery.
R&D Sandbox
The virtual sandbox is a development environment for data science, algorithm development, data analytics, and research. The environment is located in a secure cloud so researchers can benefit from cloud-based data access without compromising either information confidentiality or source organizational information infrastructure or data. The sandbox contains a variety of artificial intelligence (AI) tools and services to execute projects, conduct research, and develop data-driven products.
Examples of Data Bundles
Oncology Bundle
- Demographics
- Diagnosis
- Pathology specimens (Histology & Cytology, Staging, Mutation, Markers, Description)
- Radiation treatment
- Medication & Antineoplastic treatment
- Imaging tests (Test code and name, Interpretation text)
- Laboratory
- Outcomes
Surgical Imaging Bundle - Surgical Video
- Demographics
- Diagnosis
- Surgeries, Procedures & Anesthesia (coded data and reports)
- Perioperative complications
- Medication
- Imaging tests
- Laboratory (including Microbiology)
- Pathology specimens
- Health-related habits
- Physiological measures
Neurologic disease Bundle - Demographics
- ATD (Units and timestamps)
- Diagnosis (Coding and Urgency, Timestamps)
- Disease data: Severity, Scores and indexes, Duration
- Laboratory tests
- Imaging tests (Test code and name, Interpretation text)
- Non-Invasive Ambulatory Procedures
- Outcomes
Technology and Methodology
The I-Medata AI Center leverages a range of innovative technologies and methods to populate its data ocean. We use the collected data to develop data-driven products to boost medical care quality; to help predict and identify medical conditions as early as possible; to prioritize and optimize treatment options; to support decision-making and more.
Our center applies cutting-edge AI methodologies and technologies:
- Artificial Intelligence (AL) - Machine learning (ML) & Statistical Methods
- Deep Learning (DL)
- Cloud-based environment / On-prem
- Federated Learning - a privacy-focused machine learning method that trains models on distributed servers while preserving data privacy
- MDClone - a secure healthcare data platform that allows the analysis and sharing of large-scale healthcare data while protecting patient privacy through de-identification methods.
