Post-doctoral position: Adaptative Learning fOr Healthcare surveillAnce (ALOHA)
Keywords: healthcare, real-time analysis, machine learning, classification
Context
The goal of any health institution is to treat its patients in the best possible ways, so that their recovery progresses for the best, and as fast as possible. Within this framework and more precisely, the regional ALOHA project aims at monitoring and recording vital signs of hospitalized patients so that providing early identification of abnormalities may be possible. This work comes within automatic « track and trigger » systems.
The increase in effectiveness means a direct decrease of re-admissions and post-surgery death hazard. It induces both safety for the patients and staff efficiency for the institution. In most cases, it may even enable the patients to come back at home earlier, freeing them from the medical environment as well as liberating a bed for another incoming patient.
For this purpose, the MIS laboratory and the UPJV (Université de Picardie Jules Verne) works in collaboration with Evolucare Technologies, a software editor aiming towards the healthcare industry. In consequence, the produced work can be directly applied on software installed in healthcare institutions.
Problem
In some cases, data is given with a label informing of the patient’s situation. Otherwise, labels are not provided, and the system must locate any problem, and find a related already known problem.
In both situations, the aim is to trigger less alerts, removing the useless and noisy false-alarms, while keeping, or even enhancing, the true-positives.
Objectives
The objectives of this postdoctoral position are:
- The gathering of bibliographical content to constitute a solid basis for working on the following development.
- The development of a learning system based on real-time biomedical data, with or without labels linked to said data, through classification process. As such, a rule can be inferred to trigger alerts for the patients. In our situation, the sequential aspect of the data must be considered, and as such, notions of temporal window of data has to be included.
- The development of an enhancing system that adapts rules to patients. Each patient having a specific health condition, rules must be applicable to any patient. This way, the medical context of the patient can affect a rule, directing to a personalized medicine for each patient.
- The development of a learning system capable to predict the triggering of already learned rules, so that the healthcare staff can be alerted as early as possible.
The scientific results obtained during the postdoc will be published in conferences and journals.
Candidate profile
The candidate must hold a PhD thesis in Computer Science or Automatism, with a specialization on the Machine Learning field. He must have strong skills on the field and possibly in some frameworks and languages related to it, knowledge on the healthcare field might also help. An experience with machine learning frameworks and the Python language is strongly advised. Finally, he must have good English skills in writing and communication.
Terms of the post-doctoral position
This position begins on September 1st 2017 and will end August 31st 2018.
The monthly gross income will be of 2400 euros.
Location: Laboratoire MIS (mis.u-picardie.fr) – Quartier des Teinturiers – 14 Quai de Somme – Amiens – France
Contact:
• Jean-Luc Guérin (jean-luc.guerin@u-picardie.fr)