WITDOM use cases

WITDOM technology will be demonstrated through two project scenarios: the eHealth and the Financial Services scenarios. For each scenario several use cases were identified. This is a short description of them (read our public deliverables for further information).

eHealth scenario:

width=

  • Alignment: referred to the genomic scenario, the FASTQ file (the file needed for the analysis of the sequenced data from a patient) is automatically aligned with respect to the reference genome in the Laboratory Information Management Systems (LIMS). The genome alignment is a rather expensive computational task that thanks to the WITDOM innovations, it could be outsourced to an untrusted domain.
  • Variant annotation:
    After the alignment, the BAM file generated by the aligner is fed into a pipeline of different tools aiming at the retrieval of a list of variants in a VCF file format. VCF files contain the list of variations of the patient with respect to the reference genome. This use case requires to perform a variant annotation stage on a VCF file with respect to either a reference genome (VCF file or a public database of annotations). WITDOM’s variant annotation service is deployed in the untrusted domain and makes use of the secure signal processing, secure computation and data masking protection components in the untrusted domain.
  • Back-up: The eHealth Scenario requires backups of sensitive data (e.g. FASTQ files or other patient documents) using the untrusted domain; therefore the WITDOM platform must provide proper protection mechanisms to ensure the confidentiality and integrity of outsourced data.

Financial services scenario:

width=

  • Card fraud detection
    The main business objective this use case is to maximize the detection of fraudulent transactions while limiting the number of false positives, reducing fraud-related costs and avoiding customers’ dissatisfaction. This is a classification problem, which consists on determining which category (fraudulent or not in our case) to assign to new observations (transactions in our case), taking into account an initial and correctly classified dataset, the training set.
  • Credit risk scoring
    The “Behavioural credit risk model” is a proactive evaluation of customers’ past behaviour in order to determine a numeric value to reflect the creditworthiness of a customer.
    The "Reactive credit risk model", is a credit scoring for specific loans that answers to customers’ demand for operations with specific purposes and amount, duration and interest rate conditions. Based on this automatically generated credit scoring, business rules are applied and a suggestion is made so the credit manager or appropriate responsible (depending on the operation parameters) approves or denies the operation.
  • Cash flow
    The goal of this business process is twofold:
    1. Improved customer satisfaction: Provide the customer with an improved insight of their current and future financial situation so they can avoid undesired situations and/or plan financial actions.
    2. Marketing: Provide the bank with a more accurate view of their customers’ future situation, allowing to improve their targeted marketing, addressing customer needs with adequate anticipation (e.g. offer a loan to solvent customers temporary liquidity shortage).

The following picture shows how WITDOM architecture (core and protection components) relates to each use case, where green cells indicates which components are applicable:
width=