Chronic pain is difficult to model because no single experimental readout captures it. Sensory-neuron excitability, inflammatory signalling, behaviour and brain activity each reveal a different part of the process, while drug discovery depends on identifying preclinical signals that can predict therapeutic benefit. This challenge was the focus of an open seminar held at Łukasiewicz – PORT on 16 April 2026 by Dr. Andrii Domanskyi of Orion Pharma, Finland. His lecture was titled “Developing, characterizing, and validating in vitro and in vivo models – do we finally have precision tools for pain drug discovery?”
A recording of Dr. Domanskyi’s seminar is available on YouTube. Watch the full lecture to follow the experimental examples and methodological discussion in greater detail.
Domanskyi presented a multidimensional approach to pain research in which cellular models, behavioural assessment and neuroimaging are used as complementary layers rather than separate experiments. This is particularly relevant to analgesic development: a change measured in cultured cells may identify a mechanism, but it does not by itself show how pain-related behaviour or central processing will respond. Linking observations across levels can therefore provide a more demanding test of a candidate mechanism or treatment.
At the cellular level, the work described models based on rat and human sensory neurons, including co-cultures with rat bone marrow-derived macrophages. These systems were used to examine how specific cytokines—signalling molecules involved in inflammation—alter spontaneous sensory-neuron activity. High-density multielectrode arrays (HD-MEAs) record electrical activity from many sites at once, while chronic calcium imaging provides an optical measure related to neuronal activation over time. Both HD-MEA technology and calcium imaging were part of the experimental work covered in the recorded seminar. Together, these methods make it possible to quantify inflammatory effects functionally rather than infer them only from molecular markers.
The next level was behaviour. Pain in an animal cannot be measured by asking for a verbal rating, so preclinical research depends on observable proxies. Domanskyi discussed the BlackBox imaging platform, which uses video analysis and machine-learning algorithms to detect and quantify patterns in freely moving rodents. The approach is designed to capture spontaneous features such as movement, posture and weight distribution, providing quantitative measures that can complement conventional behavioural tests. Independent material published after the lecture confirms that Domanskyi presented results obtained with the BlackBox system in his discussion of pain research at Orion Pharma.
One validation example used rat monoarthritis models induced either by complete Freund’s adjuvant injection at the tibiotarsal joint or by monosodium iodoacetate injection into the knee. The BlackBox platform was compared with CatWalk XT for assessing static weight bearing. Both systems detected reduced weight bearing on the injured paw, and the deficit was attenuated after administration of the anti-NGF antibody tanezumab, used as a pharmacological reference in the validation. BlackBox additionally detected other behavioural parameters that responded to treatment, illustrating how automated phenotyping can extract information beyond a single predefined endpoint.
Machine learning is useful here not because it replaces biological interpretation, but because it can systematically analyse many features of spontaneous behaviour. Its value depends on whether the detected patterns are reproducible, sensitive to disease state and responsive to pharmacological intervention. In that sense, automated behavioural analysis is not simply a way to generate more data; it is a tool for increasing the resolution and objectivity of preclinical pain phenotyping.
Domanskyi also connected cellular and behavioural observations with functional magnetic resonance imaging. In a rat model of osteoarthritis, fMRI was used to examine brain activity associated with pain processing and responses to analgesic treatment. The fMRI section of the recorded seminar linked cellular and behavioural
findings to brain activity and neural patterns associated with pain and its modulation. Neuroimaging adds a system-level readout that is particularly relevant to translation because related imaging approaches can also be used in human studies, offering a potential bridge between preclinical pharmacology and clinical development.
Taken together, the seminar made a case for precision in pain drug discovery through convergence rather than through a single “best” model. Sensory-neuron activity, inflammatory cell interactions, spontaneous behaviour and brain activation each capture a different level of the pain system. When independent measures point in the same direction, researchers can make better-informed decisions about mechanisms and candidate analgesics—and identify earlier when a promising effect is restricted to one experimental context. Bringing these approaches together also creates a common ground for scientific exchange between cellular neuroscience, pharmacology, computational analysis and translational imaging.