- Research
- Emerging & Experimental
- Photobiomodulation (Light Therapy)
Does light therapy (photobiomodulation) help autistic children?
What 6 studies say — the evidence, the caveats, and what it means for your family.
Podcast·Two-voice deep dive
Listen to the discussion
Chapters & transcript
This is an AI-generated discussion of this article, transcribed automatically. The written summary below is the reviewed source — if the two differ, go by the summary.
0:00What is this discussion about?
Host A So if you buy a faulty smart speaker or like a new phone and just it won't connect to your Wi-Fi, you just, you know, you box it up and you return it. Right. It's this very simple, comforting binary.
Host B Oh, yeah.
Host A Like the tech works or it doesn't.
Host B Right. It's very black and white.
Host A Exactly.
Host B Yeah. And the distinction we have to draw today, it's paramount, really. When evaluating a device that's actively being strapped to a young child's head, the distance between what they call a promising early signal and an established proven treatment is just, it's massive.
Host A Right. Okay, so let's unpack this. Our mission for this deep dive today is to basically separate the confident sales pitches from the actual hard science. So we are looking at a stack of sources today. Specifically, it's this really exhaustive evidence synthesis protocol and findings report, and it's from Spectrum Connect. And we're diving into a therapy, it's called photobiomodulation.
Host B Yes.
Host A Which also, I think it goes by LLLT.
Host B Yeah.
Host A For low-level laser therapy.
Host B Yeah, LLLT or sometimes tPBM for transcranial photobiomodulation.
Host A Okay, got it. Transcranial photobiomodulation.
1:08What is light therapy (photobiomodulation), and how is it supposed to work?
Host A Right.
Host B So to understand this massive gap between the marketing claims and the clinical reality, we have to start with the intervention itself. Like what what is it actually doing? So photobiomodulation in this context, it involves shining a gentle near infrared or red light directly onto the brain.
Host A Onto the brain.
Host B Yeah. And the marketing literature and the clinical data we're reviewing, they often reference light specifically pulsing at a frequency of 40 hertz, utilizing a wavelength of exactly 850 nanometers.
Host A Wow. Okay, so let's actually break down what that means. Because 850 nanometers, that sounds incredibly sci-fi to anyone outside of a physics lab, I think.
Host B Yeah, it does sound very, very technical. Right.
Host A So visible red light, like we can just see with our eyes, that stops at around 700 nanometers, I think. Yeah, that's correct. But 850, that's near infrared. Yeah. Meaning it's invisible to the human eye.
Host B Exactly.
Host A But how does shining an invisible light on a child's hair or their forehead actually affect the brain inside a thick bone skull?
Host B Well, it sounds counterintuitive, right? But near infrared light actually has these really unique tissue penetrating properties.
Host A Really?
Host B Yeah. So if you've ever pressed a strong flashlight against your palm in a dark room.
Host A Oh, yeah. And it glows red.
Host B Right. Exactly. You see the red glow on the back of your hand. That's a crude version of this. Bone and tissue, they don't block near infrared wavelengths the same way they block visible light. Oh, wow.
Host A I didn't know that.
Host B Yeah, so the biological hypothesis here is that these specific photons, they basically pass through the skull and they interact directly with the mitochondria inside the brain cells.
Host A Wait, so the mitochondria, that's like the powerhouse of the cell, right?
Host B Yeah.
Host A As we all learned in high school biology.
Host B We all remember that phrase.
Host A So the light is doing what? Acting as sort of like a cellular battery charger?
Host B Well, that is the proposed mechanism anyway, yeah.
Host A Yeah.
Host B The theory suggests that there's this enzyme in the mitochondria. It's called cytochrome c oxidase. Right. And this enzyme supposedly absorbs these photons.
Host A Oh, okay.
Host B And that absorption, it kicks off a cascade that produces more cellular energy, so ATP. And it potentially reduces neuroinflammation.
Host A Okay, that makes sense on a biological level. But what about the 40 hertz pulse thing?
Host B Right. So the 40 hertz pulse, that isn't just a random number. 40 flashes per second, that actually corresponds to gamma brainwaves.
Host A Gamma waves, okay.
Host B Yeah, which are associated with high-level cognitive processing and focus. So the device manufacturers... They theorize that pulsing the light at this exact speed can entrain the brain.
Host A Entrain it.
Host B Yeah, like encouraging the brain to sync up and start producing its own gamma waves.
Host A Wow. I mean, it's a really fascinating biological hypothesis. I can completely see why a parent targeted by an ad on social media would read that explanation and think, oh, this is cutting-edge neuroscience.
Host B Absolutely. It sounds incredibly convincing.
Host A It does. Yeah. But obviously, you know, a hypothesis is not a treatment.
4:17How much research is there?
Host B No.
Host A So Spectrum Connect, they set out to find the actual hard data, right, to see if this theory actually works in practice for autism.
Host B Yes, they did. They ran a really highly rigorous search process. They mapped the available literature using something called a PRISMA flow.
Host A A PRISMA flow. Okay. What is that, exactly?
Host B Well, for those unfamiliar, PRISMA is basically it's an accounting standard for scientific literature. It tracks exactly how many studies exist globally on a topic and how they get filtered down for relevance and quality.
Host A Okay, so they use this to find all the studies on this light therapy for autism.
Host B Right.
Host A And the numbers they found, I mean, they are staggering, but not in a good way.
Host B No, not at all.
Host A They started with a discovery corpus of just 14 items across the entirety of medical literature.
Host B Yeah, 14.
Host A 14. Yeah. And then after removing duplicates and filtering out the studies that weren't actually about this specific topic... The final cut was incredibly small.
Host B Very small.
Host A Six primary studies in the entire world.
Host B Yeah. And what's fascinating here is that out of that entire global search, there are exactly zero dedicated systematic reviews for photobiomodulation in autism. Zero. We found systematic reviews for this kind of light therapy focusing on entirely different conditions, like multiple sclerosis or tinnitus depression, like traumatic brain injury.
Host A Right. Right.
Host B But for autism, it's a complete blank slate.
Host A Wow. Which, just to put that in perspective for anyone listening, a systematic review that is the gold standard of medical evidence.
Host B It really is.
Host A It's when scientists gather all the individual studies on a topic. They weigh them against each other and they say, Okay, here's the consensus. So having zero systematic reviews, it's it's basically like walking into this massive world class library, going to the front desk and asking for the definitive encyclopedia on light therapy for autism. Right. And the librarian just just looks at you, reaches into their pocket and hands you six loose handwritten index cards.
Host B Yes. That is the perfect analogy for where the science currently stands because there is no comprehensive review, right? There's no encyclopedia to rely on. So we are forced to evaluate those raw individual index cards directly to just see where these marketing claims are actually coming from.
6:32What did the main trial find?
Host A Okay, so we opened up that tiny stack of six studies, and the entire marketing apparatus basically rests on the shoulders of one single card.
Host B One card, yeah.
Host A The Fradkin 2024 study.
Host B Yes. That is the flagship piece of evidence. It is a randomized sham-controlled trial.
Host A Okay.
Host B They enrolled 30 autistic children, ages 2 to 6, and they used a specific near-infrared device called the Cognilum,
Host A over an eight week period over eight weeks okay so here's where it gets really interesting a sham controlled trial that means some kids got the real functioning device and some kids got a fake device that looked identical right exactly so they could compare the actual light therapy placebo effect to see if the light itself was doing the work Right. And the study, it actually did find a positive signal.
Host B It did.
Host A The children using the active Cognilum device, they showed a statistically significant improvement in core autism features. I think the data claims a 7.23 point change in a rating scale called the CARS-2 compared to the sham group.
Host B Yes. A 7.23 point improvement. And you know that that sounds incredibly precise and authoritative, right?
Host A It does. Yeah. It sounds very scientific.
7:46Who ran that trial, and why does it matter?
Host B But this is exactly where the Spectrum Connect audit applies the brakes and demands context. We have to look at the risk of bias.
Host A Ah, the risk of bias.
Host B Right. And for this pivotal Fradkin 2024 study, the risk of bias is officially rated as high.
Host A High. Okay, let's break down why that bias rating is so severe. Because two of the key authors on this study, De Taboada and Steingold, they are directly affiliated with JelikaLite Corp.
Host B Correct.
Host A And JelikaLite Corp is the manufacturer of the Cognilum device that was being tested.
Host B Yes.
Host A So this is basically a manufacturer-sponsored trial testing their own commercial product.
Host B Exactly. And furthermore, the primary outcome measure that that very precise 7.23 point improvement, it's based entirely on the CARS-2 scale.
Host A Okay. So what is the CARS-2 exactly? Is it like a blood test?
Host B Oh, no. It's an assessor-rated subjective scale. It is based entirely on human observation and parent interviews.
Host A Oh, wow. So it's not an objective biological marker like an MRI or something?
Host B Not at all.
Host A So we have a subjective rating scale being evaluated in a trial that's run by the people selling the device.
Host B Exactly.
Host A I mean, it's like having the CEO of a shoe company judge track meet. Right. Where some runners are wearing his brand new shoes and others are barefoot. And then he claims he's being completely objective about who looks faster.
Host B Right.
Host A Human bias is just it's impossible to untangle there.
Host B It really is. The integrity of a... subjective trial relies entirely on the blinding. Like, did the assessors know which kids were getting the real light? Right. Did the parents know?
Host A Yeah.
Host B And the audit report notes a critical flaw in sham controls for light therapy generally, which is that an active device often gives off slight warmth or subtle light cues.
Host A Oh, of course. Because it's an actual light.
Host B Yeah. So if a parent feels the helmet getting warm on their child's head, the blind is broken. They know they have the active device.
Host A Right. And then their subjective reporting on the child's behavior is just, it's inevitably going to skew positive because they think they're getting the treatment. This all ties into how the medical community grades evidence, doesn't it?
9:52Why is the evidence rated very low?
Host A I think the methodologists use something called the GRADE framework.
Host B Yes, correct. The GRADE certainty rating evaluates basically how confident we can be in the findings.
Host A Okay.
Host B Because this was a randomized controlled trial, it technically starts at a high certainty rating.
Host A It starts at high.
Host B It starts high, yeah. But it was immediately downgraded by two full points due to that high risk of bias. the manufacturer sponsorship combined with the subjective outcome. And then it was downgraded another point for something called imprecision.
Host A Imprecision. Okay, let's translate imprecision and what they call wide confidence intervals for our listeners. Because imprecision in this context, it really comes down to the size of the study, right?
Host B It's entirely.
Host A An N of 30. 30 kids, total split between the active and fake device groups.
Host B Right. So 15 kids in each group, basically.
Host A Exactly. And a wide confidence interval means the data is basically shrugging its shoulders. Like the real world result might be a massive improvement or might be almost nothing. But with only 30 kids... The math just can't focus the lens enough to tell you which one it is.
Host B Exactly. When you factor the heavy bias and the severe imprecision together, the final certainty rating for this evidence drops all the way down to very low.
Host A Very low. And yet this is the flagship study driving all these commercial claims. So, you know, a parent listening to this might logically think, Okay, the proof that it works perfectly is obviously compromised by conflicts of interest. Right. But it's just near infrared light. Like, is there any harm in just trying it?
Host B Yeah.
Host A And I think this leads us directly into the safety data. We really have to look at the rest of the evidence corpus.
Host B You do. We have to look at the other five index cards in our tiny library.
Host A Right.
11:36What do the other studies show?
Host B So we have a follow-up study, the Fradkin 2025. We have a study by the Pallanti Group in 2022. Okay. There's Ceranoglu 2022, a book chapter from Leisman in 2018, and a smaller dosing study.
Host A Okay, so let's translate some of the methodology jargon associated with these other studies so we actually understand what we're looking at here. The Fradkin 2025 study, it's described as an open-label, uncontrolled follow-up.
Host B Right.
Host A That means there is no fake device group anymore. Everyone knows they have the real thing. So the placebo effect is just, it's running wild.
Host B Totally unconstrained, yeah.
Host A And the Pallanti study is a retrospective chart review. Which just means doctors are looking backward at old patient notes to see how things went. They aren't running an active controlled experiment.
Host B Exactly. And the Ceranoglu 2022 study that focused on a small group of high functioning adults, which, as you know, is a vastly different biological and developmental population than a two year old child.
Host A Oh, yeah, entirely.
Host B So from an efficacy standpoint, none of these uncontrolled or distinct population studies can definitively prove the light itself made a difference for for day to day skills or sleep.
Host A Right. They can't prove efficacy.
12:46Is light therapy safe?
Host A But what do they tell us about safety?
Host B Well, the data across these small studies, it generally reports the therapy is well tolerated with no serious adverse events.
Host A Well tolerated. Okay. But my alarm bells kind of ring at the phrase reported as well tolerated when the sample size is this microscopic.
Host B As well, they should. I mean, the critical methodological warning here is that current safety data remains extremely limited.
Host A Yeah.
Host B The main trial had 30 children. If you have a rare but serious adverse effect, I mean, something that happens to, say, one in 100 children or one in 500, a trial of 30 children is statistically almost guaranteed to miss it entirely.
Host A Wow. Right. Because if a device has what the report politely calls a low but non-zero harm surface, which honestly is a very scientific way of saying we are actively strapping a heat and light emitting electronic device to a developing toddler's skull.
Host B Yes, exactly.
Host A A sample size of 30 is nowhere near large enough to declare it universally safe.
Host B No, it's not. If we if we connect this to the bigger picture, the safety data, it isn't showing active harm like we aren't seeing a red flashing light indicating immediate danger. Okay. But the severe lack of extensive safety data combined with this incredibly high commercial pressure to sell these devices to parents.
Host A Right. It is a massive red flag.
Host B In the classic trap, right? Right. Like selling the absence of evidence as the evidence of absence.
Host A Exactly.
Host B Saying, you know, oh, we didn't see any harm in these 15 kids in the active group. Therefore, it is perfectly safe for every toddler.
Host A Right. Which is scientifically unsound.
Host B Absolutely. And this brings us to the actual official status of this intervention according to the Spectrum Connect decision table, which is bluntly categorized as not supported, not enough research yet.
Host A Yeah. Efficacy is in doubt due to heavy bias and safety data is limited by tiny study uses.
Host B Yeah.
14:35Is an independent trial under way?
Host A But, you know, the situation isn't entirely hopeless.
Host B Okay.
Host A We need to look to the horizon to see what will finally give parents the clear answers they actually deserve. There's an ongoing trial right now registered as NCT 05363982. Okay. And this is taking place at Massachusetts General Hospital.
Host B Oh, mass general. Okay, so I'm assuming the mass general trial is finally taking the manufacturer's thumbs off the scale. Like, are they doing a truly independent double blinds study with no financial stake in the outcome? Yes. Finally, yes. And that independent factor, it changes everything. An independent trial without manufacturer ties is literally the only way to move the status from, you know, not enough research yet to an actual trustworthy conclusion.
Host A That is so important. And this is specifically for parents evaluating these devices right now. Waiting for this independent research is absolutely vital.
Host B Yes. Beautifully said. And this transition from biased data to demanding independent verification, it's actually built into the Spectrum Connect audit system itself.
Host A Oh, really?
Host B Yeah, through something called the Gate D protocol.
Host A Oh, I want to break down Gate D for our listeners because it's a brilliant piece of technology. It really is. So Spectrum Connect uses an advanced AI system to read, scrape, and summarize just vast amounts of medical literature, right? Right. So Gate D, it's basically a digital bouncer at the club.
Host B Yeah.
Host A When the AI system read that biased Fradkin study, it attempted to write a publication basically claiming, hey, light therapy works.
Host B Right, right.
Host A But the human-coated guardrails, which is Gate D, it stepped in and it just slammed the door.
Host B It did. It correctly blocked the AI-generated publication from presenting this as a settled effect. The protocol recognized that a single small manufacturer-run trial that relies on subjective outcomes, it must not be waved through to parents as a proven treatment. It demanded independent replication.
Host A I just I love that. It's like a digital immune system against bad science.
Host B That's a great way to put it. Yeah.
Host A And currently, Spectrum Connect is bridging the science of autism therapies to make it more approachable for parents, which is just it's [unclear].
Host B Absolutely.
Host A And they need rigorous, critical minds to help reinforce this bridge between, you know, raw clinical data and the kitchen table where families have to actually sit and make these heavy decisions.
Host B Yeah, it's so important.
Host A So if you are a credentialed expert, an MD, PhD, PsyD, or DO, and would like to help us in validating our findings, email us at research@spectrumconnect.us.
Host B It really is essential work, you know, cutting through the noise to protect vulnerable families from this aggressive marketing.
Host A Yeah.
17:17What are the key takeaways?
Host A So let's summarize the honest bottom line on photobiomodulation for autism.
Host B Okay.
Host A As it stands today, it is an early idea. It has a promising but incredibly fragile signal, heavily driven by a single trial with massive, massive conflicts of interest.
Host B Yes.
Host A And the safety data, it just remains limited due to very small study sizes. And right now it is being aggressively marketed to families well, well ahead of the actual evidence. We basically just have to wait for the independent trials to tell us the truth.
Host B We do. But, you know, this raises an important question for the future, though.
Host A What's that?
Host B Well, the source material makes a very careful distinction between surrogate outcomes, you know, like changes on an EEG reading and lived outcomes.
Host A Lived outcomes, meaning the actual day-to-day behavioral changes the child and the family experience.
Host B Exactly.
Host A Because fixing a squiggly line on a brain scan, it doesn't matter if the child's actual quality of life hasn't improved.
Host B Right. Exactly. But here is a really provocative thought to ponder. If those upcoming independent trials do eventually show a benefit in those subjective lived behavioral assessments, you know, like the CARS scale, how will researchers ever successfully separate the actual biological impact of the near infrared light from the incredibly powerful psychological placebo effect of family experiences?
Host A Oh, wow. I mean, that is. Yeah. Because you're handing a desperate family this glowing high-tech device.
Host B Right.
Host A And you're telling them it is cutting-edge neuroscience and it's penetrating their child's brain.
Host B Yeah. The sheer hope the family feels. Yeah. The sudden change in how the parents interact with their child simply because they deeply believe the child is being healed.
Host A Yeah.
Host B That placebo effect is monumental.
Host A It is. That is heavy. I mean, going back to where we started, when you buy a smart speaker, it doesn't work better just because you really, really hope it will.
Host B No, it doesn't.
Host A The microchips, they don't care about your feelings. But the human brain and a family's dynamic, that is a wildly different ballgame.
Host B It really is.
Host A We want the clean binary lines of an x-ray, but we're dealing with the messy, hopeful, just incredibly complex reality of human behavior.
Host B And until the science can truly untangle that invisible light from the very visible hope, we have to tread very, very carefully.
Host A Yeah. Something to mull over until next time. Keep asking questions, everyone.
Key Takeaways
One small trial found a positive signal — in a randomized, sham-controlled study, children who got real light therapy improved more on an autism rating scale than children who got a fake device.
That trial has a financial conflict — two of its authors work for the company that makes the device, and no independent lab has repeated the result.
Everything else is uncontrolled — day-to-day skills, sleep, and attention were only measured in studies with no comparison group, so we can't tell if the light made a difference.
An independent, non-manufacturer trial is already under way — worth checking back on before trying a marketed device now.
What this means for you
There's one small, early trial suggesting that gentle near-infrared light directed at the head might ease core autism traits more than a fake version of the device — a genuinely interesting result on a standard autism rating scale.
But it's a single study of 30 children, two of its authors are affiliated with the company that sells the device, and no independent lab has repeated it. Every other study lacked a comparison group, so none of them can show the light itself was responsible for anything.
Who was studied. Ages 2–6 in the main trial, using one specific device (near-infrared light, ~850nm) for about 8 weeks; a separate small study looked at adults. Generalizability beyond this narrow group is not established.
Where the studies landed
Not enough research yet1 Points toward it helping
5 Uncontrolled — can't establish cause
Tap any tile to read it. No systematic review of this therapy for autism exists yet, so every study was read on its own.
Points toward it helping1
Uncontrolled — can't establish cause5
Spectrum Connect reviews published research on interventions parents are exploring for their autistic children — so you can see where the evidence actually stands. No agenda, no selling, no cherry-picking. Just the studies, our method, and what it means for you.
See the research behind this
For anyone who wants to check our work · 6 studies
Where we looked
We built a reproducible search string for PubMed, the Cochrane Library, and Epistemonikos (below) — but found zero systematic reviews of light therapy specifically for autism. Every review we located covered a different condition (jaw disorders, MS, tinnitus, depression, brain injury) or non-invasive brain stimulation broadly. So we read the individual studies directly.
("Autistic Disorder"[Mesh] OR "Autism Spectrum Disorder"[Mesh] OR autis* OR ASD) AND ("photobiomodulation" OR "low-level laser*" OR LLLT OR "transcranial photobiomodulation" OR tPBM OR "near-infrared light" OR "red light therapy")
Run on PubMed →
What we did with what we found
What the strongest evidence says
One small sham-controlled trial found scores improved significantly more with real light (7.2-point difference, 95% CI 2.4–12.1) — but certainty is Very Low: a single trial, a subjective rating scale, and the device maker co-ran it.
Only measured in studies with no comparison group — we can't tell whether the light made any difference to these.
No serious side effects reported in the small studies so far — but this rests on very few children over short (~8-week) trials.
Keep reading
What would you like to do?
Want to talk to a parent who’s been there?
We match you with families whose kid is like yours. Free, takes two minutes.
Heard of something we haven’t covered?Suggest an article
If another parent, a clinic or a video mentioned a therapy you can’t find here, tell us. We’ll read the research and review it next.
Suggest an articleThink we got something wrong?Report a factual error
We publish the whole record so it can be checked, and that only counts if we act on what you find. If a number looks wrong, a study is missing or has been retracted, or we’ve read a finding in a way the evidence doesn’t support, tell us.
You don’t need a research background to file one. “This doesn’t match what our doctor told us” is a useful report. Every one reaches a person: we reply within seven days, and within thirty we have either corrected the page or told you when we will. Substantive reports send the affected steps back through the protocol and need fresh sign-off before anything here changes.
Report a factual errorAre you a clinician or researcher?Review this article
If you have read this and the evidence holds up, you can sign off on it. We send you a personal review link where you confirm your credentials, and our team checks every review before anything is published. Once approved, your name and a link to your profile appear on this article.
Ask to review this