The findings back the use of exercise as a primary treatment strategy for poor sleep patterns, say the researchers.

Characterized by difficulties falling and staying asleep, and early morning awakening, the prevalence of insomnia ranges from 4-22%, note the researchers. It is associated with heightened risks of various mental and physical health conditions, including dementia and cardiovascular disease.

Drug treatments for insomnia are not without their side effects, and cognitive behavioral therapy (CBT), while effective, isn't always available due to the shortage of trained therapists, explain the researchers.

An emerging body of research suggests that exercise is helpful, but current guidelines don't specify which types of exercise might be most beneficial. The researchers therefore set out to plug this knowledge gap, with a view to informing clinical practice and helping patients choose the most appropriate exercise for managing their insomnia.

They scoured research databases looking for relevant randomized clinical trials published up to April 2025 and included 22 in a network meta analysis -- a statistical technique used to simultaneously compare multiple interventions.

The trials involved 1348 participants and 13 different treatment approaches to ease insomnia, seven of which were exercise based: yoga; Tai Chi; walking or jogging; aerobic plus strength exercise; strength training alone; aerobic exercise combined with therapy; and mixed aerobic exercises. These programs ranged from 4 up to 26 weeks in length.

The other approaches included CBT; sleep hygiene; Ayurveda; acupuncture/massage; nothing; and existing treatment, such as usual care and/or lifestyle changes, the durations of which ranged from 6 to 26 weeks.

Validated scoring systems for sleep quality and insomnia severity -- PSQI and the ISI45 -- as well as subjective and objective measures of total sleep time, sleep efficiency (percentage of time spent asleep while in bed), number of awakenings after going to sleep, and time taken to fall asleep (sleep latency) were used to assess sleep patterns.

Compared with existing treatment, CBT is likely to result in a large increase in total sleep time based on subjective sleep diary data. It may also improve sleep efficiency, and shorten the amount of time spent awake after falling asleep as well as sleep latency, with sustained improvements, the findings suggest.

But some of the exercise-based interventions also seemed to be effective, when compared with existing treatment.

Yoga likely results in a large increase in total sleep time of nearly 2 hours and may improve sleep efficiency by nearly 15%. It may also reduce the amount of time spent awake after falling asleep by nearly an hour, and shorten sleep latency by around half an hour.

Walking or jogging may result in a large reduction in insomnia severity of nearly 10 points, while Tai Chi may reduce poor sleep quality scores by more than 4 points, increase total sleep time by more than 50 minutes, and reduce time spent awake after falling asleep by over half an hour. It may also shorten sleep latency by around 25 minutes.

Further in-depth analyses revealed that Tai Chi performed significantly better on all subjectively and objectively assessed outcomes than existing treatments for up to 2 years.

There are potentially plausible biological explanations for the findings, say the researchers.

With its focus on body awareness, controlled breathing, and attentional training, yoga may alter brain activity, thereby alleviating anxiety and depressive symptoms which often interfere with a good night's sleep, they suggest.

Tai Chi emphasizes breath control and physical relaxation and has been shown to decrease sympathetic nervous system activity, dampening down hyperarousal, they add. And its combination of meditative movement and mindfulness may promote emotional regulation, deactivate 'mental chatter', and reduce anxiety. It may also help to curb the production of inflammatory chemicals over longer periods, they suggest.

Walking or jogging may improve sleep by increasing energy expenditure, curbing cortisol production, improving emotional regulation, boosting secretion of the sleep hormone melatonin, and enhancing the amount of deep sleep, they continue.

The researchers acknowledge that 15 (68%) of the included trials contained design and methodological flaws. And there were no standardized, quantifiable metrics for the frequency or intensity of exercise interventions, while the sample sizes of some of the studies were small.

Nevertheless, they conclude: "The findings of this study further underscore the therapeutic potential of exercise interventions in the treatment of insomnia, suggesting that their role may extend beyond adjunctive support to serve as viable primary treatment options.

"Although current clinical guidelines make only limited mention of exercise, this study provides relatively comprehensive comparative evidence that may inform the development of more specific and actionable clinical recommendations.

"Given the advantages of exercise modalities such as yoga, Tai Chi, and walking or jogging -- including low cost, minimal side effects, and high accessibility -- these interventions are well-suited for integration into primary care and community health programs."

And there may well be one type of exercise that is best suited to easing a particular symptom of insomnia, they suggest, which further research may clarify.

Led by cognitive neuroscientist Dr Nathan Caruana, researchers from the HAVIC Lab at Flinders University asked 137 participants to complete a block-building task with a virtual partner.

They discovered that the most effective way to signal a request was through a specific gaze sequence: looking at an object, making eye contact, then looking back at the same object. This timing made people most likely to interpret the gaze as a call for help.

Dr Caruana says that identifying these key patterns in eye contact offers new insights into how we process social cues in face-to-face interactions, paving the way for smarter, more human-centered technology.

"We found that it's not just how often someone looks at you, or if they look at you last in a sequence of eye movements but the context of their eye movements that makes that behavior appear communicative and relevant," says Dr Caruana, from the College of Education, Psychology and Social Work.

"And what's fascinating is that people responded the same way whether the gaze behavior is observed from a human or a robot.

"Our findings have helped to decode one of our most instinctive behaviors and how it can be used to build better connections whether you're talking to a teammate, a robot, or someone who communicates differently.

"It aligns with our earlier work showing that the human brain is broadly tuned to see and respond to social information and that humans are primed to effectively communicate and understand robots and virtual agents if they display the non-verbal gestures we are used to navigating in our everyday interactions with other people."

The authors say the research can directly inform how we build social robots and virtual assistants that are becoming ever more ubiquitous in our schools, workplaces and homes, while also having broader implications beyond tech.

"Understanding how eye contact works could improve non-verbal communication training in high-pressure settings like sports, defense, and noisy workplaces," says Dr Caruana.

"It could also support people who rely heavily on visual cues, such as those who are hearing-impaired or autistic."

The team is now expanding the research to explore other factors that shape how we interpret gaze, such as the duration of eye contact, repeated looks, and our beliefs about who or what we are interacting with (human, AI, or computer-controlled).

The HAVIC Lab is currently conducting several applied studies exploring how humans perceive and interact with social robots in various settings, including education and manufacturing.

"These subtle signals are the building blocks of social connection," says Dr Caruana.

"By understanding them better, we can create technologies and training that help people connect more clearly and confidently."

The HAVIC Lab is affiliated with the Flinders Institute for Mental Health and Wellbeing and a founding partner of the Flinders Autism Research Initiative.

Acknowledgements: Authors were supported by an Experimental Psychology Society small grant.