Role of machine analysis of behaviors in studies on pain involving animals

The assessment of the effectiveness of analgesics is strongly based on obse rvational data from behavioural tests. These tests are interesting to give a quantization of the effect of the drugs on alive animals but their use is subjected to several difficulties : (1) many tests are difficult to analyz e as they are only based on the evaluation of a reflex response, (2) the te sts dealing with more integrated responses are also more difficult to use a nd closely depends on experimenter's subjectivity. Automation can contribut e to optimise these tests. The use of signal processing devices allows the automated measurement (and thus objective) of a pain related parameters (am plitude of a reflex, vocal emission intensity). Video image analysis allows the quantization of more complex behaviours (pain-induced specific motor b ehaviours) as well as several meaningful information during a same experime ntation (exploratory behaviour, total motor activity, food behaviour). More over, these methods make possible to get a more objective measurement, to r educe animal experimenter interactions, to ease system use, and to improve effectiveness. The prospects to work in this field are multiple : continuat ion of the attempts at an automation of the behaviours specifically induced by chronic pain; development of actual animal pain monitoring based on ana lysis of specific and non specific behavioural modifications induced by pai n. In this context, the automation of the behavioural analysis is likely to make possible real ethical progress thanks to an increase in the test's ef fectiveness and animal pain. Nevertheless, there is some limits due to tech nological problems and characteristics of the behavioural expression of pai n.