Predicting and Preventing Suicide: Psychology, Risk, and Real-Time Tech in Practice

In this Speaking of Psychology episode, Dr. Matthew Knock discusses why suicide remains difficult to predict, the most relevant risk and protective factors, and the interventions that are currently most effective. The conversation also explores how advances in technology, data analytics, and digital health tools are changing the landscape of suicide prevention, including real-time monitoring and just-in-time interventions. The episode closes with practical guidance for friends and families on how to support someone who may be at risk.

• Key insights into real-world risk assessment and decision making in clinical settings
• Overview of evidence-based therapies for suicidal thoughts and behaviors
• Role of technology, apps, and AI in prediction and prevention
• Practical steps for starting conversations and getting help

Episode overview

The podcast is a discussion with Dr. Matthew Knock, a Harvard psychologist and suicide research leader, about the persistent challenge of predicting and preventing suicide. The conversation emphasizes the need to move beyond stigma and toward scalable, evidence-based approaches that can help people in crisis. The host frames suicide as a major public health issue, noting that millions experience suicidal thoughts while actual attempts and deaths remain relatively unpredictable at the individual level.

Why predicting suicide is so hard

The episode reviews historical patterns showing that suicide rates in the United States have not substantially declined over a century, unlike other major causes of death. Dr. Knock argues that advances in other health areas came from sustained funding, campaigns, and public health infrastructure, whereas suicide has lagged due to stigma and incomplete dissemination of effective interventions. He highlights a stark contrast: while mortality from heart disease, cancers, and accidents has fallen dramatically, suicide has not, underscoring the need for a similar level of public health attention and action.

Surveys show suicidal thoughts are fairly common: globally about 9% report serious thoughts, and in the U.S. the lifetime figure is around 15%. Of those who think about suicide, roughly a third go on to attempt; globally about 3% and in the U.S. about 5% of those with such thoughts will make an attempt. Among high school students, about 20% report past-year thoughts of suicide, signaling particularly high prevalence in adolescence. However, experiencing thoughts does not predict behavior with perfect accuracy; many factors influence whether thoughts become action.

Risk factors, disorders, and predictive value

The discussion clarifies that certain risk factors such as mental illness and substance use are associated with suicide risk, but they do not reliably forecast who will attempt or die by suicide. For example, 90–95% of people who die by suicide have had a diagnosable mental illness, yet most people with mental illness do not die by suicide. Depression is consistently linked to suicidal thoughts, but other factors—especially anxiety, impulse control problems, and substance use—better predict actual attempts among those with thoughts. Across children, adolescents, adults, and veterans, mental disorders show broad consistency as predictors of thoughts and attempts, with edge differences that may help tailor risk models to specific populations (e.g., military combat exposure or family structure in adolescence).

Viewing suicide through the lens of psychological pain helps explain why people consider ending their lives. The most common reason given in qualitative interviews is an effort to escape intolerable psychological pain. This framing helps connect risk factors to the core experience of risk and can inform interventions that reduce pain and increase coping and meaning.

Clinical risk assessment and hospitalization decisions

Clinicians in settings like emergency departments face the challenge of determining who requires hospitalization. The podcast notes substantial variability in how risk is assessed in practice, including the use of standard scales as well as broader case conceptualization. Hospitalization decisions depend not only on imminent danger but also on the patient’s outpatient support, housing stability, prior treatment history, and available care networks. A recent study highlighted by Dr. Knock suggests that hospitalization benefits are not uniform: about half of patients show no clear difference with or without hospitalization, about a quarter improve after hospitalization, and about a quarter do worse. This variability underscores the need for better, more personalized pathways of care and follow-up after discharge.

Interventions: therapy, medications, and advanced treatments

Outside the hospital, evidence-based psychological interventions dominate. Cognitive therapy for suicidal behavior, a form of cognitive behavioral therapy, and dialectical behavior therapy consistently reduce suicidal behavior risk. Collaborative approaches such as CAMS (Collaborative Assessment and Management of Suicidality) emphasize understanding the patient’s drivers and tailoring treatment accordingly. In inpatient or urgent care contexts, brief, targeted versions of these approaches are being developed for rapid deployment during short hospital stays and subsequent outpatient application.

Pharmacological options include mood stabilizers like lithium for bipolar disorder, which are associated with lower suicide risk, and Clozapine for certain psychotic disorders. Fast-acting interventions are also explored in hospital settings, including ketamine and electroconvulsive therapy (ECT), with transcranial magnetic stimulation (TMS) showing promise as a noninvasive option. Research is also examining accelerated TMS protocols to shorten treatment timelines. Psychedelics are increasingly studied as potential anti-suicidal agents, though rigorous clinical guidance and safety considerations remain essential. A key point across these options is that the highest risk period for suicide often occurs in the weeks following hospital discharge, highlighting the need for robust post-discharge monitoring and transitional care akin to other medical fields like cardiology.

Technology and data driven prevention

Technology is transforming suicide risk prediction and prevention by leveraging electronic health records (EHRs) and machine learning to identify high-risk individuals within healthcare systems. These predictive models use diagnoses, medications, pain indicators, and other data to forecast risk over time, achieving increasingly accurate predictions though with challenges such as false positives. Parallel work uses smartphones and wearables to monitor mood, sleep, activity, and social connectivity, feeding into just-in-time adaptive interventions that provide support when risk rises and clinicians are not immediately available. Early work has shown that model-driven analysis of digital signals can identify many imminent risk periods, enabling targeted outreach or automated coping strategies at the moment they are needed most.

A notable line of research combines passive data collection with active participant input to forecast suicide attempts and hospitalizations up to a week in advance, using survey and sensor data. There is cautious optimism about the potential of digital interventions to complement traditional therapy, enabling timely support even when a person cannot access regular care.

Chatbots, AI, and ethical use of technology

Chatbots and large language models are discussed as tools with important potential to democratize access to evidence-based interventions and mental health resources. The guest emphasizes that technology is not inherently good or bad; the impact depends on thoughtful design, guardrails, and integration with human care. Given gaps in access to care and significant disparities in mental health equity, AI-powered tools could help bridge gaps but must be developed with rigorous safeguards to avoid harm, misinformation, and privacy concerns.

Practical advice for listeners and caregivers

The podcast closes with actionable guidance for parents, friends, and relatives. A core takeaway is that asking someone about suicidal thoughts does not implant an idea or escalate risk; asking with curiosity and care can save lives and open channels for help. The AIR framework—Ask, Show interest in the response, and Refer to professional help—offers a practical structure for conversations. If someone is at immediate risk, call emergency services or contact the 988 Lifeline. If there is no immediate danger but concern persists, help locate a clinician using trusted directories and follow up to ensure the person connects with care. The host and guest stress not bearing this burden alone; collaboration with professionals and ongoing support are essential for safety and recovery.

Concluding reflections

Throughout the talk, the emphasis is on moving from stigma and silence toward proactive, evidence-based care that respects individual needs and contexts. The podcast underscores the potential of data-driven tools to complement human judgment while acknowledging the current limits and the need for careful implementation, ongoing validation, and ethical safeguards. The overarching message is one of hope and practical action: talk openly about suicide, seek help early, and utilize the best available interventions and technologies to support vulnerable individuals at the moments they need it most.