Narrative Intelligence Studio

Medicine began as a story told by a patient to someone who listened. It has become an encounter in which the listening is the first thing cut. In a study of 112 recorded clinical visits, clinicians asked about the patient's concerns in 36 percent of encounters, and when they did, they interrupted after a median of 11 seconds. Patients who were allowed to finish took six seconds to state their concern.⁶ The story was never long. It was simply not heard.

The costs of not hearing are measurable at every scale. The World Health Organization's landmark review found that adherence to long-term therapies in developed countries averages 50 percent,⁷ which means that half of everything medicine prescribes for chronic disease is not taken, and no molecule can work in a body it never enters. Physicians are burning out: 45 percent reported at least one symptom of burnout in 2023, an improvement from 63 percent at the height of the pandemic, and still nearly twice the odds of other American workers.¹² The systematic review of narrative medicine training programs found that the outcomes participants report most often are empathy, perspective-taking, resilience, and burnout mitigation,¹³ which is to say that the skill medicine has cut is the one its practitioners say they need.

Outside the clinic, the story is being told by everyone except the people responsible for the truth of it. Ninety-six percent of American adults have encountered at least one of ten false health claims tested by KFF,¹⁴ and a third of parents believe the false claim that the MMR vaccine causes autism is probably or definitely true.¹⁵ Trust in one's own doctor, still the most trusted source, fell from 93 percent in 2023 to 85 percent in 2025, and trust in the CDC and FDA fell further.⁸ Nearly six in ten people across sixteen countries regret a health decision made on misinformation, 45 percent of young people say they disregarded their provider's advice in favor of friends or family in the past year, and young people are twice as likely as their elders to believe that someone who has "done their research" knows as much as a doctor.¹⁶ Lived experience has become a credential. That is the environment in which health stories now compete.

Into this environment has come AI. Sixty-three percent of Americans consider AI-generated health information reliable, and 65 percent of people who search online for health information have seen AI answers at the top of the results.⁹ Sixty percent say they would be uncomfortable if their own provider relied on AI in their care.¹⁷ The public will read a machine's health story and distrust a machine's medicine, and both instincts are rational. Meanwhile the digital tools built to carry health behavior are abandoned almost as soon as they are opened. Across 93 mental health apps with at least 10,000 installs, the median thirty-day retention rate was 3.3 percent.¹⁸ Adherence to apps for chronic disease averages 56 percent, and the factors that raise it are personalization, timely reminders, and "personal support complementary to the digital intervention."¹⁹ The machine needs a human story to keep a human engaged.

Three institutions hold the pieces. Medicine holds the science and has cut the story. Health marketing holds the story and is regulated for good reason: the FDA's 2025 enforcement action cited research finding that 100 percent of pharmaceutical social posts highlight benefits while 33 percent mention harms.²⁰ The tool vendor holds the generator and knows nothing about the person on the other side of the screen. Nobody holds the whole.

The name is the definition, and it is built on the companion definition of a Cinematic Intelligence Studio: an institution that makes cinematic commercial creative through a persistent, governed intelligence, accountable from brief to performance. A Narrative Intelligence Studio is that institution turned toward a different result. Performance becomes outcome. The brand becomes the person.

Narrative. The word is chosen against "content" and against "messaging." Content fills a slot. Messaging tells a person what to do. Narrative is what a person can enter, and the entering is the mechanism: the psychological literature calls it transportation, "a combination of attention, imagery, and feelings, in which an individual becomes immersed in a narrative world,"²¹ and the trials show that transported audiences change what they believe and do.

Intelligence. The word has a history in this exact territory. "Narrative intelligence" was coined by David Blair and Tom Meyer in 1997 in a paper on tools for interactive cinema, adopted by Michael Mateas and Phoebe Sengers for a 1999 symposium as "this human ability to organize experience into narrative form," and defined in their 2003 volume as "the confluence of narrative, Artificial Intelligence, and media studies."²²²³²⁴ The studio adopts the term and adds the part the academics left open: an intelligence that is specific to the people it serves, and that learns from what happens to them.

Studio. The word is chosen against "agency," "vendor," "app," and "program." A studio makes the work, owns the means of making it, and answers for the result. In health, the result is a measurable change in a person, and the accountability is regulated.

Definition

A Narrative Intelligence Studio (NIS) is an institution that produces cinematic narrative for measurable outcomes in health and wellbeing through a persistent, governed intelligence rather than through conventional production alone. It holds three capacities in one place: the craft of story in motion, held to the standard of narrative medicine; a persistent intelligence that encodes the people it serves, the craft, the record of consent, medical review, and disclosure, and the outcomes attributed to each narrative decision; and an operating model that carries a brief from clinical or brand intent through generation, review, and delivery to measured outcome, inside the rules that govern health communication. It treats generation models as interchangeable instruments. It is human-directed at every decision that matters, AI-assisted wherever assistance is possible, and controlled by the institution accountable for the person's care or trust.

What it is not

A Narrative Intelligence Studio is not a medical practice and does not deliver care. It is not therapy, though it draws on the evidence that narrative interventions relieve distress. It is not a health advertising agency, which holds the message and outsources the craft and the outcome. It is not a medical communications firm, which holds the science and rarely the story. It is not a digital health application, which holds the person's attention for a median of thirty days and loses it. And it is not a generator, which can produce a thousand health stories and cannot tell which of them is true, consented, or safe.

The studio and the Large Brand Model

The intelligence at the center of a Narrative Intelligence Studio is a Large Brand Model built for a health institution: a persistent, model-agnostic intelligence that encodes the institution's voice and visual grammar, its story world, its record of what was approved and refused and under what consent, and what each narrative decision did in the lives of the people who received it. For a hospital system the brand is the system and its clinicians. For a health-coaching application the brand is the coach. For a public health agency the brand is the agency's credibility, which the trust data show is the scarcest asset it has. The architecture is the same. The outcome it is trained on is different.

This section is the reason the paper exists, and it is written to be checked. Nothing in it claims that a story cures a disease. What the randomized evidence shows is that narrative interventions move measurable outcomes, some of them physiological, most of them psychological and behavioral, with effect sizes that range from small to large depending on the population and the outcome, and with honest failures alongside the successes.

Physiological outcomes

In 2011, Houston and colleagues randomized 299 African American patients with hypertension at an inner-city clinic to receive either three DVDs of fellow patients telling their own stories about living with high blood pressure or an attention-control DVD. Among patients whose hypertension was uncontrolled at baseline, systolic pressure fell 11.2 mm Hg further in the story group at three months and diastolic pressure fell 6.4 mm Hg further. Patients whose pressure was already controlled did not differ.² The stories did what the medication alone had not, for the patients who most needed it.

Twelve years earlier, Smyth and colleagues randomized 112 patients with asthma or rheumatoid arthritis to write for twenty minutes on three consecutive days about the most stressful experience of their lives, or about neutral topics. At four months, asthma patients who wrote about stress improved their forced expiratory volume from 63.9 to 76.3 percent of predicted; arthritis patients reduced physician-rated disease severity by 28 percent. Forty-seven percent of the writing group showed clinically relevant improvement against 24 percent of controls.³ The intervention cost a pen.

Cancer

Cancer is where the question is asked most urgently, and where the answer requires the most care. Stanton's randomized trial of expressive writing in early-stage breast cancer found significantly fewer somatic symptoms and fewer medical appointments for cancer-related morbidities at three months.²⁵ Milbury's larger trial in 277 patients with renal cell carcinoma found significantly lower symptom burden and better physical functioning at ten months, and no effect on fatigue, depression, or sleep.²⁶ A 2026 meta-analysis of eleven randomized trials of narrative interventions in cancer patients found moderate-quality evidence of large improvements in quality of life and self-efficacy and low-quality evidence of reductions in anxiety and depression, with no effect on fear of recurrence, and its authors called for "a cautious approach to clinical interpretation."⁴ A 2024 network meta-analysis of 32 trials in gastrointestinal cancer found that narrative nursing significantly reduced anxiety and depression against conventional care,⁵ and a 2025 network meta-analysis of 136 trials and 23,154 cancer patients found that narrative interventions, alongside digital cognitive behavioral therapy and virtual reality, significantly improved quality of life.²⁷

The dignity therapy trial is the caution. Chochinov randomized 441 terminally ill patients to dignity therapy, standard palliative care, or client-centered care. On the primary outcome, distress, the three groups did not differ. On the patient-reported secondary outcomes, dignity therapy was significantly more likely to be rated as helpful, to improve quality of life and sense of dignity, to change how families saw and appreciated the patient, and to lessen sadness.²⁸ A narrative intervention at the end of life did not reduce measured distress and did, by the patients' own account, change how they were seen. Both facts belong in the record.

Behavior

Kreuter randomized 489 low-income African American women to a video of breast cancer survivors' stories or an informational video with the same content. Overall, mammography at six months did not differ significantly, 49 against 40 percent. Among women without a high school diploma it doubled, 65 against 32 percent. The narrative video was better liked, enhanced recall, and reduced counterarguing.²⁹ Murphy's study of 758 women found that a narrative film about cervical cancer outperformed a non-narrative film with identical information on knowledge and attitudes, and that transportation, identification with characters, and emotion each carried part of the effect.³⁰ A recent pilot in Malawi is the honest counterweight: among women living with HIV, both a narrative video and a plain video roughly doubled cervical screening uptake over printed materials, and the narrative video did no better than the plain one.³¹ Video carried the effect. The story, in that setting, did not add to it.

What the whole literature says

The pooled estimates are sober. Frattaroli's meta-analysis of 146 randomized studies of expressive writing found a positive, significant, and small average effect.³² Braddock and Dillard's meta-analysis across 37 to 40 studies found narrative exposure associated with narrative-consistent beliefs, attitudes, and intentions at correlations of .17 to .19 and with behavior at .23.³³ Shen's meta-analysis of 25 health communication studies found a small overall effect with a decisive moderator: audio and video narratives produced significant effects, and text did not.³⁴ The World Health Organization's scoping review of more than 3,000 studies concluded that the arts have "a major role" in the prevention of ill health, the promotion of health, and the management and treatment of illness.³⁵ Kreuter's framework names the four things narrative does that information cannot: overcoming resistance, facilitating processing, providing surrogate social connection, and addressing emotional and existential issues.³⁶

Read together, the literature says three things. Narrative works best where trust and literacy are lowest, which is where the need is greatest. Narrative in motion works where narrative in text does not. And narrative is a variable with a dose, a population, and a failure mode, which makes it something an institution can be accountable for.

A studio that is accountable for outcomes has to know four things, and the health system today keeps each of them in a different place.

The studio knows the person. Who is receiving the story, what they already believe, what they fear, what they have tried. The evidence is specific about why this matters: the same survivor-story video that made no overall difference doubled screening among women without a high school diploma,²⁹ and the same storytelling DVDs that moved blood pressure in uncontrolled patients did nothing for controlled ones.² A narrative delivered to the wrong person is an expense. Delivered to the right one, it is a treatment effect. This is the knowledge medicine holds in the chart and marketing holds in the segment, and neither holds in a form a story can be built from.

The studio knows the craft. The neuroscience is unusually direct. Hasson's laboratory showed that a tightly authored Hitchcock episode produced similar brain responses across viewers in more than 65 percent of the cortex, that a Sergio Leone film produced 45 percent, an unscripted comedy 18 percent, and unstructured footage of a public square under 5 percent.³⁷ Authored narrative synchronizes audiences. The same laboratory showed that a speaker's brain couples with a listener's during successful communication and that the coupling vanishes when communication fails.³⁸ Small, Loewenstein, and Slovic showed that people gave $2.83 to an identifiable victim against $1.17 to statistical ones, and that adding statistics to the identifiable victim's story reduced giving rather than raising it.³⁹ The craft the studio encodes is the craft of making one person's story land in another person's body, and of knowing when a number will break the spell.

The studio knows the record. Every narrative carries its lineage, and in health the lineage is regulated. The HIPAA Privacy Rule requires an individual's written authorization before protected health information is used for marketing, and the authorization must disclose any remuneration.⁴⁰ The FTC's 2024 rule bans testimonials that misrepresent a real person's experience, including AI-generated ones, while permitting synthetic actors that are not presented as real patients.⁴¹ The FDA is sending thousands of letters over benefit-heavy drug promotion.²⁰ A story that moved a patient and cannot show its consent, its medical review, and its disclosure is a liability with a heartbeat.

The studio knows the outcome. Blood pressure, adherence, screening uptake, distress, quality of life, quit attempts, appointment rates. Attributed to the narrative decision that produced it, and written back, so that the next brief begins knowing that the peer story worked and the expert lecture did not, for this population, in this form.

Prediction and the Large Brand Model

Narrative medicine calls the underlying skill narrative competence, and Charon's definition is the studio's operating standard: acknowledge, absorb, interpret, act.¹ Columbia founded its program in 2000 and the first master's degree in the field in 2009.⁴² What the clinical field built as a human skill, the Large Brand Model at the center of a Narrative Intelligence Studio holds as an institutional one: persistent across staff and models, specific to the people served, and improving with every outcome it records. The prediction is bounded by the evidence in Section 4. The studio can predict that a peer story in video will outperform a text pamphlet for a low-literacy audience, because the trials say so, and it can learn where that prediction fails, because it measures.

Entertainment has been running health interventions for fifty years, and the results are among the largest in the public health record. Miguel Sabido's 1975 Mexican telenovela Ven Conmigo was followed by more than 500,000 adults enrolling in literacy classes; he calls the method "entertainment, with a proven social benefit."⁴³ Hollywood, Health & Society at the University of Southern California has spent two decades providing writers on hundreds of shows with accurate health information, funded by the CDC and major foundations.⁴⁴ East Los High, a scripted drama with digital extensions, was evaluated in the American Journal of Public Health and drew more than 215,000 tracked visits to its health resources; its evaluator's first principle was that "it has to be compelling so that you don't feel like you're being preached at."⁴⁵⁴⁶ In 2025, viewers of The Pitt were 27 percent more likely to seek organ-donation information and 39 percent more likely to seek end-of-life planning information.⁴⁷

The largest documented narrative intervention in American public health is an advertising campaign. The CDC's Tips From Former Smokers, built entirely on real former smokers telling their own stories, prompted an estimated 1.6 million quit attempts in its first three months in 2012,⁴⁸ cost $480 per smoker who quit and $268 per year of healthy life gained, fifteen times below the standard cost-effectiveness benchmark,⁴⁹ and led more than one million American adults to quit between 2012 and 2018. The CDC's own attribution: "Personal stories from former smokers about the impact smoking has had on their lives have served an important role in helping others to quit."⁵⁰ The truth campaign against youth smoking initiation reduced the risk of starting by 20 percent and prevented an estimated 450,000 adolescents from trying smoking.⁵¹

These are studio outcomes achieved without a studio: each one required a coalition of agencies, producers, scientists, and funders assembled for the purpose and dissolved afterward, taking what they had learned with them. A Narrative Intelligence Studio is the institution that keeps the coalition standing.

From brief to outcome

A brief enters a system that already knows the population, the institution's voice, and what has worked for people like these. Candidates are generated inside that grammar by whichever models best serve the scene and refused when they fall outside it. A director shapes the story; a medical reviewer holds the science; a producer holds the record. Consent is confirmed before any real person's likeness, voice, or history is used, and any synthetic person is disclosed as one. Fair balance is applied where a treatment is discussed. The work ships with provenance. The outcome returns, attributed to the narrative decisions that produced it, and the next brief begins higher.

Governance is the product

Health communication is the most regulated storytelling in the world, and a studio treats the regulation as the design brief. Written HIPAA authorization before a patient's story is used, with remuneration disclosed.⁴⁰ No testimonial that misrepresents a real person's experience, and no synthetic patient presented as real.⁴¹ Fair balance in any communication about a treatment, in an enforcement climate where the FDA has resumed sending warning letters in the thousands.²⁰ The FDA's own draft framework for AI in regulatory decisions offers the governing idea: a "risk-based credibility assessment" of an AI model "for a particular context of use."⁵² A Narrative Intelligence Studio applies the same standard to every story it generates. Credible for whom, for what decision, on what evidence.

AI-assisted, human-directed, and, in health, patient-controlled. The person whose story is told owns the consent. The institution accountable for the person's care owns the review. The studio owns the craft and the record.

On September 22, 2026, the Lucas Museum of Narrative Art opens in Exposition Park, Los Angeles: 300,000 square feet, more than 100,000 square feet of galleries, a thirteen-acre campus designed by Ma Yansong, and an inaugural presentation of more than 1,200 works drawn from a collection George Lucas assembled over fifty years.⁵³⁵⁴ The cost is reported at roughly one billion dollars, including an endowment.⁵⁵⁵⁶ It nearly did not come here. Chicago's lakefront site, announced in 2014, died in litigation in 2016; San Francisco's Presidio had refused the project earlier; Los Angeles was chosen in January 2017, and Mayor Eric Garcetti called it "an incredible gift" that "has come home."⁵⁷⁵⁵

Author's note. I was among those who made the case to George Lucas for Los Angeles as the museum's home. I claim no credit for the decision, which was his and Mellody Hobson's. I claim only that the case was right, and the argument for it is the argument of this paper: that the city which houses the studios should also house the institution that takes storytelling seriously as an art with consequences.

The museum's definition of its subject is the definition this paper has been reaching for. "Narrative art tells the story of a society, what the common beliefs are that hold it together," Lucas has said,¹¹ and more pointedly: "A picture is very powerful, and stories are very powerful. When you combine them, those two elements create a society." Narrative art "creates a group of people with a common belief system that allows them to work together, not kill each other, and do things that one person can't do."⁵⁸ The collection makes the argument in objects: Norman Rockwell and N. C. Wyeth beside Jack Kirby and Robert Crumb, Frida Kahlo and Jacob Lawrence beside Judith Baca and Diego Rivera, and the production archives of Star Wars beside all of them.¹¹ Lucas calls illustration and comics "the people's art."⁵⁴ Residents of the surrounding 90037 zip code, the museum's South Los Angeles neighbors, will enter free.⁵⁹

The connection to medicine is not decorative. The trials in Section 4 are trials of pictures combined with stories: a DVD of a neighbor with hypertension, a film about a woman getting a screening, a former smoker on television with a hole in her throat. What moved blood pressure and quit attempts was narrative art in Lucas's exact sense, a common belief system created by combining image and story, applied to a person who had not believed the statistic. A museum that treats that combination as an art worth a billion dollars and a permanent home is making, in public, the claim that a Narrative Intelligence Studio makes in practice: the craft is serious, its effects are real, and it deserves an institution.

Los Angeles is now the only city with the studios that make narrative in motion, the museum that collects it as art, and, this paper argues, the category of studio that applies it to outcomes. The three belong together.

Governed, repeatable, measurable. A hospital system, a health brand, a digital health company, or a public health agency that works with a Narrative Intelligence Studio gets one institution accountable for the story and its outcome, instead of a coalition assembled per campaign and dissolved afterward.

For hospitals and health systems

Patient experience is already money. Two percent of Medicare payments to participating hospitals are withheld and redistributed on performance, and patient experience is one of four domains that each count for a quarter of the score.⁶⁰ Hospitals with excellent patient-reported experience ran net margins of 4.7 percent against 1.8 percent for those with low ratings, and a ten-point rise in top-box ratings was associated with a 1.4 percent margin gain.⁶¹ Medical marketing by hospitals, cancer centers, and health services grew from $542 million to $2.9 billion between 1997 and 2016,⁶² almost all of it produced by the fragmenting institutions of the companion paper. A studio that holds the system's story world, its clinicians' voices, and its patients' consented narratives, and that attributes experience outcomes to what it makes, replaces a procurement problem with an asset.

For health brands and pharmaceutical companies

Direct-to-consumer prescription drug advertising exceeded $10 billion in 2024 and is shifting toward streaming and digital, where cinematic formats live.¹⁰ The enforcement climate has turned: warning letters went from one in 2023 to thousands in 2025.²⁰ The evidence says narrative in video moves belief and behavior where text does not,³⁴ and the regulation says every claim needs balance and every testimonial needs to be true. A studio built with fair balance, consent, and disclosure as checkpoints is the only form in which the evidence and the regulation are compatible at scale.

For digital health and coaching

The law of attrition, named by Eysenbach in 2005, is that "a substantial proportion of users drop out before completion or stop using the application."⁶³ The median thirty-day retention of health apps is 3.3 percent.¹⁸ The factors that raise adherence are personalization, timely prompts, and personal support.¹⁹ Video-based education improves patient knowledge in three quarters of studies and behavior in about half,⁶⁴ which argues for pairing narrative with coaching rather than replacing it. A studio that delivers a brief, cinematic, consented story at the moment a person needs it, inside the coaching relationship, is the form the attrition evidence points to.

For public health

The Tips campaign is the template and the warning. A billion-impression narrative campaign that cost $268 per healthy year of life and was run as a coalition that had to be rebuilt each cycle.⁴⁹ A studio keeps the intelligence: which stories, for which audiences, in which form, produced which quit rates, so that the next campaign begins from the last one's floor. In a trust environment where a third of parents believe a debunked vaccine claim,¹⁵ the ability to reach the "malleable middle" with a true story from a trusted face is a public good, and it needs an institution.

What the leader keeps and stops paying for

The leader keeps everything that matters: the science, the consent, the review, the direction, and the accumulated outcome intelligence, which belongs to the institution and persists across models, vendors, and staff. The leader stops paying for the coalition rebuilt each cycle, the story that reached the wrong population, the testimonial that could not show its consent, and the app that lost the patient on day thirty.

Generation is commoditizing. The craft of a story that changes a person is not.

When a health video costs nothing to generate, the institutions that sold health videos have no business, and value migrates to whatever stays scarce. Three things stay scarce in health narrative, and they are the three the trials keep pointing at.

The craft stays scarce. Audio and video narrative moves people and text does not.³⁴ Authored narrative synchronizes two-thirds of the cortex and unstructured footage synchronizes almost none of it.³⁷ The effect is strongest in the populations that need it most and vanishes in the ones that do not.²⁹² That is a craft with a dose, and dosing is a discipline.

The intelligence stays scarce. No general model knows this hospital's patients, this coach's members, or this agency's malleable middle, and no general model can be told which of its stories moved blood pressure last quarter. The knowledge of what to make, for whom, in the language of a director and the discipline of a clinician, held in a system that persists and learns from outcomes, exists nowhere today.

The accountability stays scarce. Health communication is the most regulated storytelling in the world, the public is losing trust in every source at once, and a majority now takes AI health content as reliable while a majority also distrusts AI in their own care.⁹¹⁷ The institution that can show, for every story, who consented, who reviewed, what was disclosed, and what happened next is the one that earns the trust the others are spending.

The studio form holds all three, as it held them for the Cinematic Intelligence Studio, with the outcome changed. The companion paper argued that the film industry itself is choosing the studio form for AI. The public health record argues that its greatest narrative successes were studio outcomes achieved without a studio. The category this paper names is the institution that makes those outcomes repeatable.