The science
Most people with schizophrenia who stop their medicine do so for reasons that can be seen coming: it does not seem to help, the side-effects are hard to live with, or the illness itself makes the need for treatment hard to see. Sensus measures each of those reasons, every four weeks, so they can be talked about before they become a reason to stop.
What this adds up to
People stop treatment for reasons that can be measured. Sensus measures them every four weeks, in the person’s own words and a family member’s, and puts them in front of the clinician, so the conversation about a medicine can happen before the decision to stop it.
The problem Sensus is built to help with.
74%
of 1,493 people with schizophrenia in the largest US comparison of antipsychotics stopped their assigned medicine within 18 months. The median was about six months.
CATIE trial, Lieberman et al., NEJM 20051Up to 28%
stopped because the medicine was not working well enough (15–28% across the five medicines). Up to 18% stopped because of side-effects they could not tolerate (10–18%).
CATIE trial137 studies
A systematic review found that stopping was linked to a greater risk of relapse, hospitalization and suicide.
Higashi et al., 20132The same review found the most common drivers of stopping were lack of insight into the illness, beliefs about the medicine, and substance use. What helped people stay on treatment was a good relationship with their doctor and seeing a benefit from the medicine.2 Side-effects are a further, well-documented reason: in CATIE, up to 18% stopped because of them.1
| Why people stop | What Sensus watches | How it helps |
|---|---|---|
| “It isn’t helping” | Senses, negative symptoms, function | Shows the person, in their own answers, whether things are improving, so benefit can be seen, or its absence raised early |
| Side-effects | Body: side-effects, pain, sleep, weight | Puts what the medicine is costing the person in front of the prescriber, in detail, before it becomes the reason to quit |
| Lack of insight | Awareness | Tracks how the person sees their illness and their treatment, the driver most often reported |
| Substance use | Context | Keeps cannabis and other substances in view, as both a reason for stopping and a cause of worsening |
| Mood and distress | Feelings | Depression and anxiety are common alongside schizophrenia and add to the burden of treatment |
| The relationship with the doctor | The person’s report and the clinician’s report | Gives both sides the same record to talk from, so the visit starts from what has actually happened |
Why “Sensus”
Sensus is Latin for sense, feeling, perception and understanding, which is what the platform captures, so the person, their family and their clinician share one understanding of how treatment is going.31
Who it is for, what happens at each sitting, who answers what, and where the answers go.
Adults aged 18 and over who have a diagnosis of schizophrenia or a related psychotic disorder and are being treated. Sensus does not test whether someone has schizophrenia.
The person answers most of each sitting. A family member or partner answers the everyday-function questions separately. The clinician adds their own ratings and test results at visits. The three accounts are never blended.
A core sitting every four weeks, about 20 minutes. Every third sitting adds a quarterly module, about 15 minutes more. A trial of a medicine is judged over at least six weeks,27 so every trial holds several readings.
The full set is taken at the start so there is a baseline for every part, then the quarterly module returns every twelve weeks.
| Sitting | Part | Assessment | Answered by | Items |
|---|---|---|---|---|
| Core every 4 weeks | Senses | Colorado Symptom Index | Person | 14 |
| Feelings | M3 Checklist | Person | 27 | |
| Body | GASS side-effects | Person | 22 | |
| Body | DVPRS pain | Person | 5 | |
| Body | PROMIS Sleep 8a | Person | 8 | |
| Treatment | Adherence questions | Person | 2–3 | |
| Context | Home or elsewhere | Person | 1 | |
| Quarterly added every 12 weeks | Function | WHODAS 2.0, 12-item | Person and family member | 12 + 12 |
| Negative symptoms | MAP-SR | Person | 15 | |
| Awareness | VAGUS-SR | Person | 10 | |
| Context | TAPS substance use | Person | about 4 | |
| Senses | R-GPTS paranoia, where paranoia is the main concern | Person | 18 | |
| At visits | All | Clinician’s own ratings, cognition scores and test results | Clinician | — |
Every instrument Sensus gives is published and validated. Each card gives what it measures, who answers, what the research shows, and whether it has been shown to work when repeated over time.
How to read the evidence
Sensitivity is the share of people who truly have a condition that a screen correctly flags. Specificity is the share without it that the screen correctly clears. AUC runs from 0.5 (no better than chance) to 1.0 (perfect). Reliability (alpha or ICC) shows how consistently a scale measures; 0.70 or above is generally considered acceptable.
Each card also carries a label for repeated use, because Sensus is a monitor:
What it does. Asks how often in the past month the person has had a range of psychiatric symptoms, including hearing voices, feeling suspicious, and thoughts of harming themselves or others.
What the research shows. In 3,874 adults receiving Medicaid, a score of 30 separated people with psychiatric disabilities with a sensitivity of 76% and a specificity of 68%. Internal consistency 0.92; test-retest 0.71.4 It was given at baseline, 6, 12, 18 and 24 months in a Toronto trial of 378 people with serious mental illness, a quarter of whom had a psychotic disorder.5
How it is scored. Each of the 14 items asks how often a symptom occurred in the past month: not at all (0), once (1), several times during the month (2), several times a week (3), at least every day (4). Items are summed to a total of 0–56; higher means more distress. A total of 30 or more is the published clinical cut-off.
What it does. Measures ideas of reference (feeling that events refer to oneself) and persecutory thoughts (feeling that others intend harm), with published severity bands.
What the research shows. Built from more than 2,000 patients with psychosis. A persecution score of 11 distinguished 360 patients with persecutory delusions from 7,297 people without, with a sensitivity of 92.8%, a specificity of 85.2% and an AUC of 0.953. A score of 18 correctly identified 81% while wrongly flagging only 7%. Reliability above 0.90 across the range of severity.6 Used as a repeated outcome from baseline to 24 weeks in a randomized trial of virtual-reality therapy.7
How it is scored. Each item is rated 0 (not at all) to 4 (totally). Part A, ideas of reference, has 8 items (0–32); Part B, persecution, has 10 (0–40). Published severity bands for Part B: average 0–5, elevated 6–10, moderately severe 11–17, severe 18–27, very severe 28 and above. For Part A: average 0–9, elevated 10–15, moderately severe 16–20, severe 21–24, very severe 25 and above.
What it does. Measures the loss of motivation and pleasure, the “negative” symptoms that are hard to treat and strongly shape long-term recovery, and that a person may not think to mention.
What the research shows. Internal consistency 0.90 in people with schizophrenia or schizoaffective disorder, and a correlation of 0.65 with the clinician-rated measure it was built to mirror.8 Two-week test-retest 0.63.9
How it is scored. Each of the 15 items is rated 0–4 and summed to 0–60. Higher scores mean more motivation and pleasure, so a falling score means negative symptoms are worsening. No clinical cut-off is published; Sensus reads it against the person’s own baseline.
What it does. Measures four kinds of awareness: that experiences are symptoms, that there is an illness, that treatment is needed, and what the illness costs. Lack of insight is the reason for stopping treatment most often reported.2
What the research shows. Validated in 215 people with schizophrenia-spectrum disorders and correlated strongly with established insight measures.10 Two-week test-retest 0.92 in a later validation.11 In a sham-controlled trial, VAGUS scores improved with treatment and the improvement held for four weeks.12
How it is scored. Each of the 10 statements is rated from 0 (strongly disagree) to 10 (strongly agree), with 5 meaning unsure. Items are averaged within each of the four kinds of awareness, and the total is the average of those four. Higher scores mean greater insight.
What it does. The World Health Organization’s measure of everyday functioning: understanding and communicating, getting around, self-care, getting along with people, life activities and taking part in community life.
What the research shows. Internal consistency 0.86 and test-retest 0.98 across more than 65,000 respondents.13 In 352 people with schizophrenia: 0.94 and 0.92.14 The 12-item version: 0.89 in 881 people with psychosis.15 In serious mental illness the family-member version was more sensitive to change than the person’s own,16 which is why Sensus asks both.
How it is scored. Each of the 12 items asks how much difficulty there has been in the past 30 days: none (0), mild (1), moderate (2), severe (3), extreme or cannot do (4). The simple score is the sum, 0–48; the WHO’s complex scoring converts it to 0–100. Higher means more difficulty. The person’s and the family member’s scores are shown side by side, never averaged.
What it does. Asks about the side-effects of antipsychotic medicines: sleepiness, stiffness and restlessness, dry mouth, weight, sexual and hormonal effects, and how much each one bothers the person.
What the research shows. Against a clinician’s side-effect rating in 81 outpatients, most of them with schizophrenia, 86% of GASS items had a sensitivity above 75% and 64% had a specificity above 75%.17 One-week test-retest 0.80; internal consistency 0.81–0.84.18
How it is scored. Items 1–20 ask how often a side-effect occurred in the past week: never (0), once (1), a few times (2), every day (3). Items 21–22 are scored yes (3) or no (0). The total gives three published bands: 0–12 absent or mild, 13–26 moderate, 27 and above severe. A separate column records which effects the person finds distressing; it is not scored, but it is shown to the clinician.
What it does. Pain now, with words under every number, then how much pain has interfered with sleep, activity, mood and stress. About a third of people with schizophrenia live with clinical pain, and it is easily missed.19
What the research shows. Correlated 0.93 with the standard 0–10 pain scale in 350 patients.20 Internal consistency 0.87.21 Scores moved clearly with pain after surgery.22
How it is scored. Pain now is rated 0–10, with a description under every number (6 is “hard to ignore, avoid usual activities”). Four further items rate, 0–10, how much pain interfered with activity, sleep, mood and stress over the past 24 hours. Each is shown on its own; there is no single total.
What it does. Sleep quality, depth and restfulness over the past week. Sleep problems are common in schizophrenia and are a side-effect of several medicines.
What the research shows. Reliability of 0.90 or above across most of the range, and more precise than the Pittsburgh Sleep Quality Index.23 In people with psychosis, internal consistency 0.91, and judged suitable for this group.24
How it is scored. Each of the 8 items is rated 1–5 for the past seven days, and the raw total (8–40) is converted to a T-score, where 50 is the US average and every 10 points is one standard deviation. Higher means more sleep disturbance. HealthMeasures’ general guidance treats 55 and below as within normal limits, 55–60 as mild, 60–70 as moderate and above 70 as severe.
What it does. Reads depression, anxiety, the signs of bipolar disorder and post-traumatic stress in one sitting, along with how much they interfere with daily life. One item asks about thoughts of suicide, and answering it opens the 988 crisis pathway whatever every other score says.
What the research shows. In 647 adults in primary care, compared against a structured diagnostic interview, the M3 Checklist detected any of the four disorders with a sensitivity of 83% and a specificity of 76%. By module: depression 84% / 80%, bipolar 88% / 70%, anxiety 82% / 78%, PTSD 88% / 76%.3
How it is scored. Each of the 27 items is rated on a five-point frequency scale, from not at all to most of the time. The items give a total score and module scores for depression, anxiety, bipolar signs and PTSD, plus four questions on how much symptoms interfere with daily life. For those individuals acknowledging any trouble with their mood state, a total of 33 or greater identifies 83% of true cases of a diagnosable condition. The suicide item is read on its own, whatever the total. M3’s weighting and band tables are proprietary.
What it does. Screens for use of tobacco, alcohol, cannabis, prescription medicines used other than as prescribed, and other drugs.
What the research shows. In 2,000 adults in primary care, it detected problem use with a sensitivity of 93% for tobacco, 82% for cannabis and 74% for alcohol, with specificity of 79–87% for tobacco and alcohol and 93% or higher for other substances. The authors note further refinement is needed before it is recommended as a screen for substance use disorder.25
How it is scored. Part 1 asks how often in the past year the person has used tobacco, drunk heavily, used drugs, or used prescription medicines other than as prescribed. Any use opens Part 2: two or three yes/no questions for each substance about the past three months. Each substance scores 0–3: 1 or more indicates problem use, and 2 or more a higher risk of a substance use disorder.
What it does. Diagnosis, remission and treatment resistance are defined on ratings that need a trained clinician.2728 Sensus holds a place for each result, so the clinic’s gold standard and the home record sit on one timeline.
| Instrument | What it measures | Who gives it | Rights |
|---|---|---|---|
| SCID-5 | Structured interview confirming a DSM-5 diagnosis and ruling out others | Trained interviewer | Licensed (APA Publishing) |
| Mental Status Examination | Appearance, behaviour, speech, thought, perception, insight, judgement | Clinician | A clinical method |
| PANSS | Positive, negative and general symptoms, 30 items; defines remission and response | Trained rater | Licensed (Mapi Research Trust) |
| CAINS | Negative symptoms: motivation, pleasure and expression | Clinician interview | Free for research (Kring lab) |
| Calgary Depression Scale | Depression, separated from negative symptoms, 9 items | Clinician interview | Free for clinicians in routine practice |
| BACS or MCCB | Cognition; the MCCB is the outcome standard for cognition trials29 | Trained tester | Licensed |
| Laboratory work | CBC, TSH, liver and renal panels, vitamin levels; metabolic monitoring; ANC on clozapine; toxicology | Clinician | Standard tests |
What the research shows. In a 2025 study at three sites with 56 people with early-course schizophrenia, smartphone versions of the Trail Making Test and Symbol Digit Substitution correlated with the MCCB composite at r = 0.60 and r = 0.53. Only three of ten tasks reached moderate test-retest reliability.30 An earlier study of the smartphone Trail Making Test found no practice effect over three months of weekly use.26 These results are promising, but they do not yet match the clinic standard, so any home cognition reading would be labelled exploratory.
What they do. Short questions no published instrument covers, chosen because they are what helps people stay with treatment.2
| Question set | When | What it records |
|---|---|---|
| Background | Once, at the start | Year symptoms began, hospital stays, family history of psychosis or bipolar disorder, other conditions, current medicines (no doses) |
| Adherence | Every four weeks | How much of the medicine was taken (all, most, about half, a little, none) and what got in the way: forgetting, side-effects, not feeling it was needed, not seeing it help, cost or refills |
| Home or elsewhere | Every four weeks | Where the person is answering from, after Ranjan et al. 2022 |
| Life events | Every four weeks | Work or study, time with family or friends, a move, a stressful event, missed doses, stopping a medicine, an ER visit, a hospital stay |
| Personal goals | Named once; rated every four weeks | One to three goals in the person’s own words, each rated 0–10 |
How it is scored. Adherence below 80% is flagged first on the report, because no conclusion about a medicine can be drawn while doses are being missed. Life events and goals are not scored; they are shown on the timeline beside the scores.
| Assessment | Part | When | Answered by | Headline accuracy or reliability | Repeated use | In schizophrenia |
|---|---|---|---|---|---|---|
| CSI | Senses | Core | Person | Cut-off 30: 76% / 68%; alpha 0.92 | Long-term studies | Serious mental illness |
| R-GPTS | Senses | Quarterly (optional) | Person | Cut-off 11: 92.8% / 85.2%; AUC 0.953 | Treatment trial | Yes |
| MAP-SR | Negative symptoms | Quarterly | Person | Alpha 0.90; r = 0.65 with clinician measure | Not yet established | Yes |
| VAGUS-SR | Awareness | Quarterly | Person | Test-retest 0.92 | Shown to detect change | Yes |
| WHODAS 2.0 | Function | Quarterly | Person and family | Alpha 0.94, test-retest 0.92 in schizophrenia | Shown to detect change | Yes |
| GASS | Body | Core | Person | 86% of items >75% sensitivity vs clinician | Not yet established | Yes |
| DVPRS | Body | Core | Person | r = 0.93 with standard pain scale | Shown to detect change | Not yet |
| PROMIS Sleep 8a | Body | Core | Person | Reliability ≥ 0.90; alpha 0.91 in psychosis | Not yet established | Yes |
| M3 Checklist | Feelings | Core | Person | Any disorder: sensitivity 83%, specificity 76% | Not yet established | Primary care |
| TAPS | Context | Quarterly | Person | Cannabis problem use: sensitivity 82% | Not yet established | Primary care |
What the gaps mean. Several instruments have not yet been studied for repeated use, and some have not been studied in schizophrenia. Sensus reads every score against the person’s own earlier scores, and says on every report which instrument a reading came from, so a clinician can weigh each one accordingly. Where Sensus collects repeated data, it adds to the evidence that does not yet exist.
No instrument is used on this platform until written permission is in hand, and any instrument whose permission is still being sought is marked pending here and everywhere else it appears. The M3 Checklist belongs to M3 Information. The DVPRS is a US federal work, used unaltered. The TAPS Tool is in the public domain. The CSI and the GASS are listed in the PhenX Toolkit as freely available, with permission not required; written confirmation is being sought. Permission has been requested, or will be, for every other instrument on this page.
This page is information, not a diagnosis. If you or someone you know is in danger right now, call or text 988 (Suicide & Crisis Lifeline), or call 911 in an emergency.
© 2026 M-3 Information LLC. Sensus™ is a trademark of M-3 Information LLC. All rights reserved. Third-party instruments and medicine names belong to their owners.