Outracing champion Gran Turismo drivers with deep reinforcement learning.

Many potential applications of artificial intelligence involve making real-time decisions in physical systems while interacting with humans. Automobile racing represents an extreme example of these conditions; drivers must execute complex tactical manoeuvres to pass or block opponents while operating their vehicles at their traction limits1. Racing simulations, such as the PlayStation game Gran Turismo, faithfully reproduce the non-linear control challenges of real race cars while also encapsulating the complex multi-agent interactions. Here we describe how we trained agents for Gran Turismo that can compete with the world's best e-sports drivers. We combine state-of-the-art, model-free, deep reinforcement learning algorithms with mixed-scenario training to learn an integrated control policy that combines exceptional speed with impressive tactics. In addition, we construct a reward function that enables the agent to be competitive while adhering to racing's important, but under-specified, sportsmanship rules. We demonstrate the capabilities of our agent, Gran Turismo Sophy, by winning a head-to-head competition against four of the world's best Gran Turismo drivers. By describing how we trained championship-level racers, we demonstrate the possibilities and challenges of using these techniques to control complex dynamical systems in domains where agents must respect imprecisely defined human norms.

Residency match interview scheduling: quantifying the applicant experience.

BACKGROUND: The process of offering and scheduling residency interviews varies widely among programmes. Applicants report distress and have advocated for reform. However, there is a paucity of quantitative data to characterise applicant concerns. OBJECTIVE: We quantified the interview scheduling experience for US allopathic medical students in the 2020 main residency match. METHODS: An anonymous, 13 question survey was sent to student representatives from each Association of American Medical Colleges member institution. Recipients were asked to forward the survey to their entire fourth-year class. RESULTS: Of 4314 applicants to whom the survey was sent, 786 (18.2%) responded. Overall, 20.4% reported missing the opportunity to interview at a programme because they did not have adequate time to respond to an invitation; applicants into surgical specialties were significantly more likely than their non-surgical peers to report this experience (26.4% vs 18.4%, p<0.05). Most (57.4%) respondents scheduled an interview knowing they would likely cancel it in the future. The most commonly cited reason for this behaviour was concern that applicants would not receive invitations from other programmes (85.6%). A majority (56.4%) of respondents did not believe the match interview process functions based on equity and merit. CONCLUSIONS: About one in five respondents missed the opportunity to interview at a programme because they did not respond to an invitation in time. Most respondents scheduled interviews knowing that they were likely to cancel them in the future. Standardisation of the interview invitation timeline would address these concerns.

Towards gender-affirming nutrition assessment: a case series of adult transgender men with distinct nutrition considerations.

No guidelines exist regarding nutrition assessment for transgender or gender non-conforming patients. Multiple nutrition assessment methods utilize gender-specific values that provide distinct recommendations for males and females. This clinical case series depicts the food and nutrition considerations of ten adult transgender men using anthropometric, survey, and dietary recall data. Male reference values were used to analyze patient data, though multiple approaches to nutrition assessment in the transgender population are discussed. Major nutrition-related concerns were obesity, low fruit and vegetable intake and high sodium intake; disordered eating was not a prominent concern. Further research is needed to inform nutrition care for the transgender and gender non-conforming populations.

Determining Ideal Management for Patients With Coexisting Prolactinomas and Psychiatric Symptoms: A Systematic Review.

OBJECTIVE: Prolactinomas-pituitary tumors that overproduce prolactin-can cause various troublesome symptoms. Dopamine agonists (DAs) reduce prolactin production in the prolactin pathway, making them the first-line treatment for prolactinomas. However, the main side effect of DA treatment, hyperdopaminergia, is an explicit etiology for psychiatric side effects. Psychiatric conditions are often treated with dopamine antagonists, which can induce hyperprolactinemia. This presents a challenge for patients with both a prolactinoma and a preexisting psychiatric condition, as treatment of one condition could worsen the other. This review seeks to identify an adequate therapeutic regimen for patients with coexisting prolactinomas and psychiatric symptoms. METHODS: This review examined PubMed citations from 1960 to 2023 published in English and involving human subjects. Case reports, case series, and cohort studies involving patients with concomitant prolactinomas and psychiatric symptoms, as validated by brain imaging, serologic prolactin levels, and medical history or chart reports of psychiatric symptoms, were included. RESULTS: Thematic analysis included 23 reports involving 42 participants; 27 of the 42 patients experienced a significant reduction in prolactin levels and psychiatric symptoms (64%). Treatment of those 42 patients included discontinuing or altering antipsychotic/dopamine antagonist therapy or discontinuing DA therapy to reduce psychiatric symptoms, with surgery or radiation postpharmacotherapy as a last-line strategy. However, in some cases (reported in Tables 2 to 4), either psychiatric or prolactin-related symptoms recurred despite adjustment. CONCLUSIONS: Clinicians may find it beneficial to prioritize specific antipsychotics (aripiprazole, olanzapine, ziprasidone, or clozapine) over others (risperidone, thioridazine, thiothixene, and remoxipride). Discontinuing DA medication at least periodically until the patient's condition improves may also be advisable. If these 2 initial approaches do not yield a significant improvement in symptom management, surgery or radiation therapy may be considered. As patients may respond differently to these therapies, our study still recommends a patient-centered approach.

The Effect of Antipsychotics on Prolactinoma Growth: A Radiological and Serological Analysis.

Many antipsychotic (AP) medications work by reducing dopamine levels. As hyperdopaminergia is known to cause psychosis, antipsychotics work to relieve these symptoms by antagonizing dopamine receptors and lowering dopamine levels. Dopamine is also a known negative modulator of the prolactin pathway, which allows for drug agents like dopamine agonists (DAs) to be incredibly effective in managing tumors that secrete excess prolactin (prolactinomas). While the effects of DAs on prolactinoma size and growth have been studied for decades, the effects of APs on prolactinoma size remain to be seen. We hope to investigate the effects of APs on prolactinomas by conducting a thorough PubMed search, including patients with diagnosed prolactinoma on concurrent AP therapy. Our search led to 27 studies with a total of 32 patients. We identified themes regarding seven antipsychotics: risperidone, haloperidol, amisulpride, thioridazine, aripiprazole, olanzapine, and clozapine. Risperidone, haloperidol, amisulpride, and thioridazine caused a significant increase in prolactin in most cases where they were used, and prolactin decreased after their discontinuation. For example, risperidone discontinuation resulted in a decrease in prolactin levels by an average of 66%, while haloperidol, amisulpride, and thioridazine discontinuation lowered prolactin by an average of 82%, 72%, and 89.7%, respectively. However, there were some exceptions in regard to risperidone, haloperidol, and thioridazine, where prolactin levels were not as severely affected. Aripiprazole, olanzapine, and clozapine all had significant reductions in prolactin levels when patients were switched from another antipsychotic, such as risperidone or haloperidol. The average percent decrease in prolactin when switched to aripiprazole was 67.65%, while it was 54.16% and 68% for olanzapine and clozapine, respectively. The effect of individual antipsychotics on prolactinoma size was difficult to ascertain, as imaging was not obtained (or indicated) after every antipsychotic switch, and many patients were taking dopamine agonists concurrently. Therefore, it would be difficult to ascertain which factor affected size more. Also, some patients received surgery or radiotherapy, which completely negated our ability to make any assertions about the effects of certain pharmacological agents. Although it is difficult to ascertain the role that antipsychotic medications play in the formation of prolactinoma, we have found that the cessation of certain antipsychotic medications may lead to a reduction in prolactin levels and possibly the presence of a measurable prolactinoma.