Can hooking your head up to a fancy battery (tDCS) cure depression? Artwork

PsyDactic

A resource for psychiatrists and other medical or behavioral health professionals interested in exploring the neuroscientific basis of psychiatric disorders, psychopharmacology, neuromodulation, and other psychiatric interventions, as well as discussions of pseudoscience, Bayesian reasoning, ethics, the history of psychiatry, and human psychology in general.

This podcast is not medical advice. It strives to be science communication. Dr. O'Leary is a skeptical thinker who often questions what we think we know. He hopes to open more conversations about what we don't know we don't know.

Find transcripts with show-notes and references on each episodes dedicated page at psydactic.buzzsprout.com.

You can leave feedback at https://www.psydactic.com.

The visual companions, when available, can be found at https://youtube.com/@PsyDactic.

All Episodes

PsyDactic

Can hooking your head up to a fancy battery (tDCS) cure depression?

January 05, 2026 • Episode 79

In December 2025, the FDA authorized the Flow F100, an innovative at-home wearable headset that utilizes transcranial direct current stimulation (tDCS) to treat major depressive disorder. Unlike traditional pharmaceuticals that act systemically, this device targets the dorsolateral prefrontal cortex with localized electricity to modulate neuronal excitability and address the asymmetry hypothesis of depression. While the Empower study that evaluated this technology demonstrated statistically significant improvements in response and remission rates, the FDA approved it with a moderate level of uncertainty regarding its true efficacy due to a clinically insignificant 2.3-point difference on the average improvement using the Hamilton scale and potential unblinding bias in the trial. By contrasting frequentist and Bayesian statistical frameworks, Dr. O'Leary encourages a skeptical but curious evaluation of whether this technology represents a genuine clinical breakthrough or a temporary trend in the long history of electrotherapy.

Please leave feedback at https://www.psydactic.com or send any comments to feedback@psydactic.com.

References and readings (when available) are posted at the end of each episode transcript, located at psydactic.buzzsprout.com. All opinions expressed in this podcast are exclusively those of the person speaking and should not be confused with the opinions of anyone else. We reserve the right to be wrong. Nothing in this podcast should be treated as individual medical advice.

Welcome to PsyDactic. Today is Monday, Jan 5, 2026 and I am Dr. O’Leary, a psychiatrist and host of this is a podcast about psychiatry and neuroscience and occasionally other things, like game theory or ultrasound or mental health policy. Today I am going to continue talking about what happens when we apply electromagnetic energy to the head. I’m not talking about ECT, or electroconvulsive therapy. I am also not talking about TMS, or transcranial magnetic stimulation. I am talking about transcranial direct current stimulation, which is basically plugging your head into a battery and adjusting the amperage. Sounds fun, huh? When I was a kid, I remember touching the poles of a 9 volt battery to my tongue and feeling the tingling sensation. I was also dumb enough once to let a friend fool me into grabbing two ends of wires attached to each electrode and getting a little shock. Fortunately, direct current is generally safer than alternating current (which is what you get if you plug directly into an outlet in your house) and it didn’t kill me. Doing that with an outlet could easily have killed me. I am going to talk about transcranial direct current stimulation today because the FDA just approved a device that you can wear on your head at home that delivers 2 mA of current in the hopes that it lifts “the black mood,” the "melancholy," more recently known as major depressive disorder.

I need to warn you that my opinions are my own and no one else's. Please do not confuse them with those of the Federal Government, the Department of Defence, the Defence Health agency, or The Montpellier Vitalists. I also use Ai to assist me with my research and content creation, but in the end, what I give you here is my responsibility and it is my fault if it is wrong or biased, though I do a lot of fact checking and try to be fair.

In any case, I will stop jabbering and get into the content.

The following is an outline for the script of the Episode.*******

The First Home Device to Treat Depression

On December 8, 2025, The U.S. FDA granted authorization for the Flow FL-100, a wearable headset designed to treat Major Depressive Disorder (MDD) in the comfort of a patient’s home.This approval is a victory for transcranial direct current stimulation (tDCS) and one of the first machines that could potentially challenge pharmacotherapy as the primary first-line intervention for mood disorders. For decades, the standard of care has revolved around the modulation of neurotransmitters—serotonin, norepinephrine, and dopamine—through systemic chemical agents. While effective for millions, many patients fail to achieve remission with their first or second antidepressant, and a significant proportion discontinue treatment due to side effects ranging from weight gain to sexual dysfunction and emotional blunting.

The Flow FL-100 uses direct current stimulation (which is basically the same thing as hooking a battery up to your forehead and plugging it into your scalp). The electricity acts locally rather than systemically; and for depression, it targets the dorsolateral prefrontal cortex (DLPFC) with a weak direct current to hopefully change the excitability of the neurons in the brain just below where the electrodes are placed.

To fully appreciate the significance of tDCS, let's rewind the clock a little bit. Adding electricity to the human brain is not a modern invention. It is a practice that predates the discovery of the neuron, the lightbulb, and even the battery itself.

The earliest recorded instances of "electrotherapy" date back to over three thousand years BCE. Egyptians likely applied electric catfish to the body to relieve pain. This practice of using electric marine life, such as electric sting rays, was documented by the Ancient Greeks. Scribonius Largus, the court physician to the Roman Emperor Claudius, documented a peculiar treatment for intractable headaches and gout: the application of a live torpedo fish (an electric ray) to the affected area. Largus noted that when the fish was placed on the forehead of a patient suffering from a headache, the resulting numbness—a phenomenon we now recognize as stimulation-induced analgesia—could provide lasting relief.

The idea that electricity was a primary component of life became popular during the Enlightenment. In the late 18th century, Italian physician Luigi Galvani conducted his famous experiments on frogs. By connecting the sciatic nerve of a dissected frog to a metal conductor during a lightning storm (and later with bimetallic arcs), Galvani induced muscle contractions. He concluded that biological tissue contained "intrinsic animal electricity," a vital force that governed movement and sensation. Before this Descartes thought that the nervous system was hydraulic because that was a technology he could imagine. Now the experiments with electricity could be done, everyone started to try to explain movement and motivation as being electric.

Galvani's contemporary and rival, Alessandro Volta, argued that the electricity was not intrinsic to the animal but generated by the contact of dissimilar metals—a debate that led Volta to invent the voltaic pile, the world's first battery, and then Galvani's nephew, Giovanni Aldini, in 1801, began treating patients suffering from "melancholia" (a historical name for what we would likely recognize as major depression) using his uncle's principles and Volta's battery. Aldini applied direct current to the scalps of his patients, reporting that the stimulation could lift the "black mood" of the afflicted. He also like to get attention by stimulated the corpses of executed criminals in London, causing their eyes to open and jaws to clench, much to the horror and fascination of the public—and it wasn’t long after that Mary Shelley wrote Frankenstein, using the idea that electricity provided a vital force to material tissues.

The Era of Shock and Stigma

Despite Aldini's early rabble rousing, the use of electricity in medicine fell into disrepute during the 19th century, often associated with quackery. "Medical galvanism" devices promised to cure everything from baldness to impotence. It wasn't until the 1930s that electricity returned to mainstream psychiatry

In 1938, Italian neurologists Ugo Cerletti and Lucio Bini introduced Electroconvulsive Therapy (ECT). If you are interested in learning about this in more detail, my first 3 full episodes of this podcast deal with this. Unlike the subtle modulation of Aldini's direct currents, ECT utilized alternating currents at high intensities to intentionally induce a generalized grand mal seizure. The therapy was incredibly effective for severe, catatonic depression, mania, and psychosis caused by mood disorders. It isn’t the electricity itself, but its ability to produce a seizure is the therapeutic agent; the electricity is merely the trigger.

In its early days, ECT was administered without anesthesia or muscle relaxants, leading to bone fractures from the violent convulsions. It also caused significant memory loss and post-ictal confusion. Modern Anesthesia and better control of current and electrode placement has made ECT far more safe, but culturally, ECT became synonymous with "shock therapy," a stigma reinforced by media portrayals like One Flew Over the Cuckoo's Nest and also by the fact that it was vastly overused for all sorts of conditions, from bedwetting to homosexuality, when it first became popular. It has made a resurgence, but there is still a huge stigma associated with its use in society.

The Modern Resurgence: Polarization and Plasticity

The rehabilitation of more gentle forms of electrical stimulation began in the 1960s. Researchers in animal physiology, such as Bindman et al. (1964), observed that applying weak direct currents (brain polarization) to the cerebral cortex of rats could produce long-lasting changes in neuronal firing rates. These currents were too weak to trigger action potentials (firing) directly. They didn’t make neurons fire. Instead, they altered the resting membrane potential of the neurons.

Anodal (Positive) Stimulation: Was found to depolarize the membrane, bringing it closer to the firing threshold. This made neurons more likely to fire in response to natural inputs—a state of increased excitability.
Cathodal (Negative) Stimulation: Was found to hyperpolarize the membrane, pushing it further from the threshold and inhibiting activity.

In the last 30 years or so, Direct Current Stimulation, and Transcranial Magnetic Stimulation, have been researched. Magnetic stimulation uses changing magnetic fields to stimulate neurons and it was shown that applying positive direct current stimulation could make it easier for TMS to make neurons fire. It was shown that tDCS can induce a kind of neuroplasticity called Long-Term Potentiation (LTP) in the human brain non-invasively and without the need for a sophisticated magnetic field generator. Unlike ECT, there was no seizure and no loss of consciousness. The subject generally feels only a mild tingling or stinging sensation.

TMS and tDCS made to possible to affect the brain without "shocking it" but instead gently "nudging" it. It established the biological plausibility of tDCS as a treatment for conditions characterized by cortical hypoactivity, such as the left prefrontal cortex in depression. This 1998-2000 era laid the scientific bedrock upon which Flow Neuroscience would eventually build its platform.

The Flow FL-100: Genesis and Architecture

Flow Neuroscience, and its founders, Daniel Mansson and Erik Rehn made it their goal to transition tDCS from a laboratory curiosity to a regulated home-use product. Flow Neuroscience was established in 2016 in Malmö, Sweden. Daniel Mansson is a clinical psychologist who treated patients with MDD. He was familiary with the limitations of the standard of care: psychotherapy (which was effective but scarce and slow) and pharmacotherapy (which is easier to access but often accompanied by side effects like weight gain and sexual dysfunction). Many patients are also too depressed to really engage fully in therapy, and unable to tolerate the side effects of SSRIs. TMS or ECT require patients to make multiple trips to the office often daily or multiple days per week for many weeks. We need something effective that can be done and home.

Erik Rehn, the other co-founder, brought more technical expertise. A neuroscientist and electrical engineer with a background in artificial intelligence and computational biology at the Royal Institute of Technology (KTH) in Stockholm, Rehn was able to miniaturize the bulky tDCS stimulators used in research labs. The biggest challenge for a device that can be used at home was to make it foolproof—one that could ensure consistent electrode placement and safety without a technician present.

The Asymmetry Hypothesis and Device Engineering

The design of the Flow FL-100 is grounded in the "Asymmetry Hypothesis" of depression, a neurophysiological theory supported by decades of EEG and fMRI research.This hypothesis posits that MDD is characterized by a specific imbalance in the frontal lobes:

Hypoactivity in the Left Dorsolateral Prefrontal Cortex (DLPFC): This region is critical for executive function, emotional regulation, and "approach" behaviors. In depressed patients, it is often metabolically underactive, contributing to apathy, cognitive fog, and low mood.
Hyperactivity in the Right DLPFC: This region is often associated with "withdrawal" behaviors, negative affect, and anxiety. Overactivity here is linked to rumination and the inability to disengage from negative thoughts.

The FL-100 utilizes a bifrontal montage to address this dual pathology.

The Anode (+): Is placed over the left DLPFC (F3 position). It delivers excitatory current, depolarizing the neurons and encouraging LTP to restore activity to this underactive region.
The Cathode (-): Is placed over the right DLPFC (F4 position). It delivers inhibitory current, hyperpolarizing the neurons to dampen the overactivity.

The device utilizes sponge electrodes soaked in saline solution to ensure conductivity and minimize skin irritation. A critical safety feature is the impedance check; the device will not start stimulation unless it detects a solid electrical connection with the skin, preventing "hot spots" that could cause burns.

The headset connects via Bluetooth to a smartphone application that serves multiple critical functions:

Safety and Compliance: The app controls the stimulation. The user cannot simply put the headset on and turn it on; they must initiate the session through the app. The app restricts usage to the prescribed protocol (one session per day), preventing overdose or misuse.
Augmented Guidance: To solve the problem of electrode placement (a major source of error in tDCS), the app uses video tutorials and potentially camera-based guidance to ensure the headset is positioned correctly on the DLPFC.
Behavioral Therapy: During the 30-minute stimulation sessions, the app delivers behavioral therapy modules. These are based on standard CBT principles and focus on four pillars: Sleep, Nutrition, Exercise, and Meditation. This combinatorial approach—biological stimulation plus psychological skill-building—mimics the "gold standard" of combining drugs with therapy.

The Clinician Dashboard:

For the US market, the device is prescription-only. The system includes a portal that allows the prescribing psychiatrist or physician to monitor the patient's adherence. The clinician can see if the patient is skipping sessions, monitor their self-reported symptom scores (PHQ-9 or MADRS) tracked in the app, and adjust the protocol if necessary. This digital tether was a crucial factor in convincing the FDA that an at-home Class III device could be managed safely.

Class III Approval

The FDA approval of the Flow FL-100 was a complex undertaking that highlights the divergent regulatory philosophies governing drugs versus medical devices. Understanding this pathway is essential for interpreting the strength of the evidence.

Understanding the FDA Class System

The FDA classifies medical devices into three categories based on risk and regulatory control:

Class I (Low Risk): Dental floss, bandages. Minimal regulation.
Class II (Moderate Risk): Condoms, powered wheelchairs, most medical software. These usually require 510(k) clearance, where the manufacturer must prove the device is "substantially equivalent" to a product already on the market (a predicate).
Class III (High Risk/Novel): Implantable pacemakers, replacement heart valves, and novel devices that sustain life or prevent impairment. These require Premarket Approval (PMA).

Because the Flow FL-100 was the first at-home tDCS device indicated for depression, there was no "predicate device" to which it could claim equivalence. Consequently, it automatically defaulted to Class III, the most rigorous category. This meant Flow Neuroscience could not simply show their device was electrically safe; they had to prove, from scratch, that it was effective in treating MDD using valid scientific evidence.

The Breakthrough Device Designation

A critical accelerator in this process was the Breakthrough Device Designation, granted on May 31, 2022.The FDA reserves this status for devices that meet two strict criteria:

Life-Threatening or Debilitating Condition: The FDA recognized MDD as a debilitating condition with a high burden of morbidity and mortality (suicide risk).
Unmet Clinical Need/Advantage: Unlike antidepressants, which flood the entire body with substances that can have side effects, tDCS is non-systemic. It affects the brain but spares the liver, gut, kidneys, and reproductive system. Additionally, it addressed the access barrier of clinic-based TMS.

This designation does not lower the evidence standard, but it prioritized the review. It allowed Flow Neuroscience to engage in "interactive communication" with the FDA during the design of the Empower trial. This meant they could agree beforehand on the endpoints, the blinding strategy, and the statistical analysis plan, making it more likely the final application would be approved (as long as the end-points were hit).

Drug vs. Device: A Regulatory Divergence

To the lay observer, "FDA Approved" sounds the same whether it is for a pill or a headset. However, the evidentiary standards are fundamentally different.

Replication vs. "Reasonable Assurance":
Drugs (NDA): The standard is "Substantial Evidence" of efficacy. This almost always requires two independent, randomized Phase 3 trials (often with N=500+ each) to prove that the result was not a statistical fluke. The FDA assumes that biology is messy and requires replication.
Devices (PMA): The standard is "Reasonable Assurance" of safety and effectiveness. The FDA acknowledges that device trials are harder to run (blinding is difficult, prototypes are expensive). Therefore, they often accept one robust pivotal trial (like Empower, N=174) if it is well-designed and supported by real-world evidence and bench testing.
The Control Group Problem:
Drugs: Placebo pills are indistinguishable from active drugs. Blinding is perfect. If patients guess their assignment, the trial is often considered failed.
Devices: Perfect blinding is physically impossible. Active tDCS causes skin sensations (tingling). A sham device that does nothing feels different. The FDA accepts a "moderate level of uncertainty" regarding unblinding in device trials, provided the safety profile is pristine and the effect size is robust enough to outweigh the placebo effect.

This context is vital: The Flow FL-100 was approved with a smaller N (174) and more uncertainty about blinding than a new antidepressant would be. However, this is balanced by a safety profile that is vastly superior to any drug.

The Empower Study: Deconstructing the Evidence

The "valid scientific evidence" underpinning the PMA approval came primarily from the Empower Study (NCT05202119). This was a randomized, double-blind, sham-controlled, multi-center trial conducted in the US and UK.

Trial Design in a Remote World

The Empower study was unique in that it was a fully decentralized trial. Participants did not visit a clinic for daily treatments. Instead, they were mailed the device and instructed to use it at home, with supervision provided via video calls.

Sample: N=174 adults with moderate to severe MDD (mean HDRS-17 score ~19).
Design: 1:1 Randomization to Active tDCS vs. Sham tDCS.
Protocol: An "Activation Phase" of 5 sessions/week for 3 weeks, followed by a "Strengthening Phase" of 3 sessions/week for 7 weeks. Total duration: 10 weeks.
Adjunctive Use: Crucially, patients were allowed to stay on stable doses of antidepressants. This makes the results applicable to the real-world population of "partial responders," rather than just treatment-naïve patients.

The "Active Sham" Conundrum

To maintain the blind, Flow utilized an "Active Sham" protocol.

Active Group: Received 2.0 mA of continuous current for 30 minutes.
Sham Group: The device ramped up to 1.0 mA over 30 seconds (causing the initial tingling/itching sensation), then ramped down to 0 mA. It remained off for 29 minutes, then ramped up again briefly at the end to mimic the "disconnect" sensation.

This design aimed to fool the sensory nerves of the scalp. However, the FDA's analysis revealed a flaw: Unblinding.

Active Group Guessing: 77.6% correctly guessed they were on the active treatment.
Sham Group Guessing: 59.3% correctly guessed they were on the sham.

Did patients guess correctly because the sham felt different? Or did they guess active treatment correctly because they felt their depression lifting (functional unblinding)? This ambiguity is the "moderate uncertainty" the FDA flagged in its decision.

Efficacy Data: The Continuous vs. Binary Split

The results of the Empower study show some statistical tension. On the one hand, the device appears highly effective. On the other hand, the clinic significance of the findings is questionable.

Continuous Outcomes (The Averages):

The primary endpoint was the change in the Hamilton Depression Rating Scale (HDRS-17) score at Week 10.

Active: Improved by 9.41 points.
Sham: Improved by 7.14 points.
Difference: 2.27 points (p=0.012).

While statistically significant, a 2.3-point difference is modest. To put this in perspective, the FDA generally looks for a drug-placebo difference of 2-3 points. So, Flow performed on par with oral antidepressants.However, the sham response (7.14 points) was very high, likely due to the intensive interaction of the remote supervision (the "care effect").

Binary Outcomes (The Responders):

The secondary endpoints, which measured "success rates," were far more striking.

Response Rate (>50% improvement): Active 58.3% vs. Sham 37.8% (+20.5% difference; p=0.017).
Remission Rate (Symptom Free): Active 44.9% vs. Sham 21.8% (+23.1% difference; p=0.004).

The Key Insight: The device effectively doubled the remission rate of the sham. The Number Needed to Treat (NNT) for response is approximately 5. This means for every 5 patients treated, one will respond who would not have responded to placebo. In the world of antidepressants, where NNTs often hover around 7-9, an NNT of 5 is excellent.

Safety and Tolerability Profile

The safety data from Empower was unequivocal: tDCS is safer than drugs.

Serious Adverse Events (SAEs): Zero device-related SAEs. No seizures, no mania, no suicidality.
Common Side Effects: The most common issues were dermatological. Skin redness was reported by 63.5% of the active group (vs. 18.5% sham). Skin irritation and mild stinging were also common.
The Burn Issue: Two participants in the active group reported "burns." Investigation revealed this was due to the reuse of dried-out sponge pads, which increases electrical impedance and creates heat. This finding led to strict labeling: pads must be single-use and properly moistened. The FDA deemed this a manageable labeling issue rather than a fundamental flaw.

Comparing this to the side effect profile of SSRIs (insomnia, weight gain, sexual dysfunction, withdrawal tremors), the FL-100 offers a vastly superior safety proposition.

The Statistical Deep Dive: Significance vs. Utility

The Empower study results—modest average improvement but high remission rates—force a confrontation with two core statistical concepts: Statistical Significance and Clinical Significance.

The P-Value and Statistical Significance

The study achieved a p-value of 0.012 for the primary endpoint.In the frequentist framework, this means that if the device had no effect (null hypothesis), there is only a 1.2% chance of seeing a result this positive as long as there was no bias or problem with randomization in the study. Reliability: We can be relatively confident the result is not a fluke.

The Minimal Clinically Important Difference (MCID) aka Clinical Significance

However, Reliability is not the same as Utility. The concept of MCID asks: "Is the difference large enough for the patient to notice?" For the HDRS-17 scale, the consensus MCID is typically 3 to 5 points.

The Conflict: The observed difference was 2.27 points.
The Interpretation: The average patient experienced a benefit that was statistically real but arguably below the threshold of "feeling significantly better" than placebo on average.
It may be that despite there being a highly significant difference in the numbers, few of the patients or their families would actually notice a difference in their symptoms.

Metaphor: The Weight Loss Drug

Imagine a diet pill trial.

Group A (Pill): Loses 10.5 lbs.
Group B (Placebo): Loses 10.0 lbs.
Result: The difference is 0.5 lbs. With a massive sample size, this could be statistically significant (p < 0.01). But is it clinically important? No. No patient cares about losing an extra 0.5 lbs.

The question is how meaning are those 2.3 points

The "Cliff Effect": Reconciling the Data

How do we reconcile a "gray zone" average difference (2.3 points) with a "slam dunk" remission difference (23%)? This is the Cliff Effect.

Depression scores in a population generally follow a Bell-shaped Curve, also called a gaussian distribution. There are a lot more values near the center of the curve, where most people are, and a lot fewer values near the edges of that curve. Most people are closer to average than they are to the fringes. The threshold for "Remission" (Score < 7) cuts through the densely populated center of that curve. A small shift in the curve pushes many more people over an arbitrarily drawn line near the center of the curve than a line drawn near the fringes.

The Shift: The placebo effect shifted the whole curve to the left (improvement).
The Nudge: The active device shifted the curve slightly further to the left (by 2.3 points).
The Result: Because the curve is so steep in the middle, that small 2.3-point "nudge" pushed a large percentage of people over the "Remission Cliff."

The Bayesian Perspective

How might a Bayesian Framework interpret this data differently. The current standard (Frequentist statistics) relies on p-values, which only tell us how rare the data is if we assume perfect randomness and no bias. Bayesian statistics, however, deals in the Probability of Truth.

Philosophies of Probability

Frequentist (The Standard): "Probability is the frequency of an event over infinite trials." It assumes the parameter (true efficacy) is fixed and the data is random. It asks: How unlikely is this data if the drug does nothing?.
Bayesian (The Challenger): "Probability is a degree of belief." It assumes the data is fixed (we observed it) and the parameter (true efficacy) is a distribution of probabilities. It asks: Given this data, what is the probability the drug works?.

The Power of Priors: Incorporating Skepticism

The FDA cited "conflicting results from the literature" as a reason for uncertainty.A Frequentist p-value ignores this history; it treats the Empower trial as if it exists in a vacuum.

A Bayesian analysis allows us to incorporate this history using a Prior.

Skeptical Prior: We could start the analysis with a mathematical distribution that assumes tDCS is likely ineffective (based on past failed trials like the New England Journal of Medicine studies).
- Evidence for this could include the fact that prior studies that were similar to this one found no evidence of effectiveness. (https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(23)00640-2/abstract)
- There are also studies that show tDCS has little to no effect on healthy adults for a range of cognitive measures (https://pubmed.ncbi.nlm.nih.gov/25448853/#:~:text=A%20systematic%20review%20published%20in%20*Neuropsychologia*%20in,reliable%20neurophysiological%20effect%20beyond%20MEP%20amplitude%20modulation.)
- There has never been a dose response established for tDCS for depression in the past.
The Update: We then feed the Empower data (Likelihood) into this Prior.
The Posterior: If the Empower data is strong enough (which it appears to be), it can overcome the Skeptical Prior. If it is borderline, then it is unlikely to change the prior probability that tDCS is only mildly effective if effective at all for depression.

The Metaphor: The Courtroom

Frequentist Judge: Looks only at the evidence from this crime scene. "Given the crime scene, there is a 90% chance he's guilty?"
Bayesian Judge: Looks at the evidence plus the defendant's criminal record (Prior) and other circumstantial evidence. "Given he has 10 prior arrests for something similar AND given the evidence at the scene the probability of guilt is 99.9%."
- However, given the high improbability that the defendant could have been at the scene, given that he was admitted to a hospital for an appendicitis at the time of the crime, the probability that he committed the crime drops to less than 1%.

Re-interpreting Empower through Bayesian Reasoning

A Bayesian re-analysis of Empower will yield a more nuanced and (I believe) useful conclusion for clinicians, than simply relying on a single industry sponsored study of relatively few patients:

Probability of Superiority: A frequentist might say, given the adequacy of the trial and the p-value of 0.012, it is highly likely that the device has an effect and that effect is likely to be within the 95% confidence interval. A Bayesian would then say: "Bull mularky".
Adjusting for priors as well as unblinding and other biases in the study: Bayesian hierarchical models could treat the "unblinding guess" as a variable. By modeling the correlation between "guessing active" and "guessing sham.”
Adjusting for priors: After this adjusting for the data itself, then adjustments would be made based on prior data. For example, prior studies designed like this one have had negative results, so the prior probability that this study will be positive is something less than 50%. How much less would be determined by the results of prior studies.

In essence, while Frequentist stats gave us a "Yes/No" (Significant/Not Significant), Bayesian stats would give us a "Probability Profile"—a much more useful tool for clinical decision-making.

Resources