PhD. Candidate – Machine Learning and Artificial Intelligence for Forensic Non-Human DNA Recognition and Interpretation

This is a job offer for a Doctoral Candidate (DC) for European Training Network “Natural Traces” (Project 101120165, Horizon Europe, see https://cordis.europa.eu/project/id/101120165). The DC will be engaged in cutting-edge research in the context of the use of non-human DNA in forensic recognition and interpretation. This field that has seen significant research and debate in recent years. The European Network of Forensic Science Institutes (ENFSI) has published recommendations in favor of a Bayesian decision framework for communicating results to the court, which involves interpreting forensic findings through the lens of likelihood ratios (LR). This approach aims to establish a common methodology across forensic disciplines, fostering a more standardized and quantitative presentation of results to aid fact finders in legal proceedings. Despite the success of LR methodologies in disciplines such as DNA analysis, challenges persist in obtaining adequate LR values for various real-life scenarios found when non-human DNA is involved. Factors like data scarcity, multiple hypotheses, high dimensionality, and mixtures of trace materials makes the general problem challenging. Moreover, artificial intelligence and machine learning offer a great opportunity to effectively extract evidence that will need to be calibrated in the form of a LR for presentation in court. Some members of the network, including the Ph. D. Advisor for this DC, are members of associate members of ENFSI, which will ensure adequate and useful integration of the outcomes of the project for the benefit of European forensic science.

Hypotheses and Plan: The primary objective of this position is to develop and implement the tools necessary for the accurate computation and validation of likelihood ratios across diverse forensic disciplines. This can be achieved using artificial intelligence (AI), machine learning and statistical methods. The DC will collaborate closely with other sites and fellows within the ETN to achieve the following objectives:

1. Robust Computation of Likelihood Ratios: Develop methodologies for the reliable calculation of likelihood ratios with non-human DNA evidence, ensuring compatibility with published ENFSI Guideline recommendations, and facilitating interpretation within a common forensic context.
2. Validation Methodology: Establish rigorous validation procedures to assess the readiness of calculated LR values for use in casework, addressing issues such as data quality, model robustness, and statistical significance.
3. Solutions for Challenging Scenarios: Devise innovative solutions to compute LR values for complex scenarios, including mixtures of biological materials, by leveraging advanced statistical and AI techniques, and collaboration with domain experts.

Expected Progress: The successful candidate will contribute to the advancement of forensic science by achieving the following milestones:

• Development of methods to calculate LR values for non-human DNA biological trace evidence.
• Implementation of validation methodologies to ensure the reliability and validity of computed LR values.
• Collaboration with ETN members to exchange data, simulate cases, propose robust models, and validate findings.
• Dissemination of research outcomes through publications, presentations, and participation in relevant conferences and workshops.