AI-driven Platform for the Generation of Predictive Tools from OMICs Data
Empowering Biotech and Biomedical Research with a user-friendly, code-free AI solution to generate predictive tools.
About
The Digital Phenomics Platform is an online tool that enables researchers and biomedical companies to build and use predictive models for the classification of phenotypes from “omics” data. Digital Phenomics Platform uses state-of-the-art Machine Learning and Artificial intelligence for finding combinations of biomarkers with bespoke predictive power. Users can upload their “omics” data, construct datasets, build predictive models, analyse models and generate predictions in a user-friendly and code-free manner.
Features
Dataset Generation - Easy Built Your Dataset
The platform contains a tool (Data2Set) that enables the users to construct datasets with their omics data and associate them with a user-defined phenotype. The construction of datasets is simplified with a drag-and-drop functionality for omics data files (e.g. transcriptomics, metagenomics, or proteomics ) of individual samples. Users simply place the files into Positive or Negative outcomes related to a given phenotype and submit the upload. Users can add and edit information regarding the data/phenotype and save it to the dataset. The AI system will then build and safely store the dataset into a binary that the user can use anytime to train and validate predictive models. the performance based on the entire dataset and selects the bespoke model. For all bespoke models, a model description and performance evaluation are generated by the AI together with the mathematical formula for scoring the combinations of th biomarkers.
Features
Model Generation - Automated Development of your predictive model
The platform contains a feature that enables users to build predictive models for the classification of phenotypes from omics datasets. Model building is fully automated through our powerful AI engine (O2Pmgen) that applies Machine Learning to omics data. Users only need to choose the dataset and specify how much percentage of false positive rate is acceptable for your model. The AI engine will use up to 50% of your dataset for learning patterns in data and evolve models towards finding bespoke combinations of biomarkers with optimal sensitivity and specificity. The AI uses a state-of-the-art evolution-based algorithm to build multiple models, calculates the performance based on the entire dataset and selects the bespoke model. For all bespoke models, a model description and performance evaluation are generated by the AI together with the mathematical formula for scoring the combinations of th biomarkers.
Features
Model Analysis and Comparison Toolkit
Once models have been generated using our O2Pmgen AI engine, users can further evaluate and compare the resulting bespoke models through the following visualisation functionalities:
- Dynamic plotting of Receiver Operator Characteristic (ROC) curves using our ROCplot tool. This tool generates a plot that enables users to evaluate the predictive power of models.
- Evaluate the distribution of false positives and negatives across the model scoring using our FDRanalyser tool. This tool generates a plot that enables users to understand and compare the confidence associated with the model’s predictions.
- Users can select and visually compare multiple models on a panel. Plots can be downloaded as an Excel file or image in various formats.
Features
Data Predictions - Generate Prediction with models
Users can make predictions on new data using our O2Pheno tool. Users only need to select a model, input the biomarkers values and submit to the Digital Phenomics AI-engine. In a fraction of seconds the predictive score is computed and a phenotype prediction is inferred (positive or negative). The platform generated a color scale that enables the users to understand the confidence of the prediction. Saving the prediction data is also available as an option (limited to the user’s plan). Users can further consult, manage and delete their saved predictions.
Welcome to our platform! Type your E-Mail to get started and choose your plan.
| FEATURES | STARTUP | PROFESSIONAL | PREMIUM |
|---|---|---|---|
| Upload of datasets | Max 100 files | Max 1000 files | Max 10000 files |
| Maximum storage of datasets | 5 | 10 | 20 |
| Maximum storage of models | 10 | 100 | 1000 |
| Maximum storage of predictions | 100 | 1000 | 10000 |
| AI engine | Basic | Boosted | Extra-Boosted |