SIGNAL 01
How you sound
Share a short voice note in English or Urdu. We listen for changes in vocal patterns, not the words themselves.
Up to 60 seconds · Nothing is savedEARLY SIGNALS, HUMAN CARE
A quiet, private way to notice changes in how you're feeling. MindHx brings together your voice, words, and a short clinical questionnaire.
PRIVATE SESSION CONTEXT
Only age range, optional gender, relationship status, and life context are used in memory for this session. Nothing is saved as an account.
SIGNAL 01
Share a short voice note in English or Urdu. We listen for changes in vocal patterns, not the words themselves.
Up to 60 seconds · Nothing is savedSIGNAL 02
Write a little about how things have been lately. Be as brief or open as feels right.
Sent for analysis only - never saved to a database.SIGNAL 03 · 0 / 3 COMPLETE
The PHQ-9 is a validated questionnaire used by healthcare professionals. Think about the last two weeks.
PHQ9 · 1 / 9
PHQ9 · 2 / 9
PHQ9 · 3 / 9
PHQ9 · 4 / 9
PHQ9 · 5 / 9
PHQ9 · 6 / 9
PHQ9 · 7 / 9
PHQ9 · 8 / 9
PHQ9 · 9 / 9
LIVE ESTIMATE
Complete your check-in to see how the signals come together.
ABOUT MINDHX
Depression, anxiety, and related conditions are common, but the first conversation about them rarely feels easy to start. In much of Pakistan and beyond, seeking help is sometimes read as weakness or private shame - something to manage quietly rather than an ordinary health matter to bring to a professional. Add limited access to psychiatrists and psychologists outside major cities, long waitlists even where care exists, and a lot of people never take a first screening step - not because they don't want support, but because the step itself feels too large to start.
MindHx exists to make that step smaller. It replaces “I should probably talk to someone eventually” with a private, five-minute check-in you can do from a phone or laptop tonight - no appointment, no waiting room, no one else needing to know unless you choose to tell them. It is not a diagnosis and it does not pretend to be one. It is a starting point: something concrete to notice about yourself, and, if it matters, something concrete to bring into a real conversation with a real clinician.
Most self-assessments ask one kind of question and stop there - a mood questionnaire, a chatbot, or a wearable's guess from your heart rate. MindHx instead treats a check-in as three separate signals that don't always agree: how you sound (pause patterns, loudness variability, speaking pace), what you actually say in your own words rather than a multiple-choice answer, and how you score on PHQ-9, GAD-7, and K10 - the same validated questionnaires clinicians already use in practice.
Each signal is scored on its own first, then combined into one weighted estimate. The combination isn't a black box: the results page shows exactly how much each signal contributed to the final number, using an additive model where each term's share is the real, mathematically exact contribution - not an approximate explanation added after the fact. If your voice sounded flat but your questionnaire answers were mild, or the other way around, you'll see that tension laid out honestly, not smoothed away inside a single tidy number.
A screening tool that asks about your inner life only earns trust if it's honest about what happens to what you share. MindHx's default is to keep almost nothing. The core check-in requires no account at all - no email, no password, no profile to create. A voice recording is processed for acoustic features (pause ratio, pitch variability, speaking rate) and then discarded immediately; the audio itself is never written to a server or a database, and never listened to by a person. Typed text is sent for a sentiment read and then dropped - not logged, not retained, and never reused to train anything.
If you choose to create an optional account, the only thing that persists across visits is a check-in's score, risk band, and detected themes - never your transcript, your typed words, or your individual questionnaire answers. That distinction is deliberate: a therapist you eventually sit down with should hear your story from you, in your own words and your own time, not have it pre-written by an app before you arrive.
A risk score by itself doesn't help anyone - what matters is what happens after it's shown to you. MindHx routes based on what it finds, not just what it scores. If PHQ-9's self-harm item or clear crisis language shows up anywhere in a session, everything else stops immediately and the Emergency Support page opens before any score is even computed - that check cannot be bypassed by continuing the conversation.
For everything else, the results page pairs your combined signal with a support plan matched to the themes it detected - grounding techniques for anxiety, small behavioral-activation steps for low motivation, pacing guidance after loss or trauma - alongside a bounded AI chat that answers orienting questions like “what does CBT actually involve” or “is it normal to feel this way on this medication” from a fixed, clinician-reviewed reference library, never improvising medical advice of its own.
When a real conversation with a professional is the right next step, MindHx tries to make that concrete too, rather than leaving it as vague advice: a city-by-city directory of verified psychiatric and psychological care in Pakistan, and a plain answer to what to actually say at a first appointment.
MindHx is bilingual by construction, not by afterthought: every page - the check-in itself, the AI chat, the therapist directory, even this paragraph - exists in both English and Urdu, switchable with a single tap in the header, with the entire layout correctly mirroring for Urdu's right-to-left script rather than just swapping words inside a left-to-right frame. That distinction matters in a country where a screening tool available only in English quietly excludes most of the people who might actually need it.
Being built with that context also means being honest about what isn't finished yet. The combined risk score is a weighted heuristic, not a clinically calibrated probability - it hasn't been validated against real outcome data, and pretending otherwise would make it less trustworthy, not more. The Urdu translation of PHQ-9, GAD-7, and K10 is a careful draft for this session, not a licensed clinical instrument. The acoustic voice signal is an explicit heuristic proxy, deliberately built to be swapped for a real biomarker vendor later. None of that lives in fine print - it's stated plainly here and in the project's technical documentation, because a screening tool that oversells its own certainty is more dangerous than one that admits what it doesn't yet know.
MindHx is built for someone who has noticed something is off - a lower mood, more worry than usual, sleep that isn't restoring them the way it used to - and would rather understand that a little before deciding what, if anything, to do next. It's meant to be a first look, not a last resort and not a running log to obsess over daily.
It is explicitly not built for a mental-health emergency. If you or someone you're with may be in immediate danger, a local crisis line and MindHx's own Emergency Support page matter far more than any questionnaire score, and the app is built to get out of the way and point there directly the moment it detects that situation - before, not after, showing you a number. It's also not a substitute for ongoing care: for anyone already working with a therapist or psychiatrist, MindHx is at most a way to notice patterns between appointments, never a reason to skip one.
ⓘ MindHx is a screening and triage aid, not a diagnosis. Your results are a starting point for a conversation with a qualified professional. Brand ↗