# SPUN Medika — Medical Tourism Price Estimator

**Design plan · PRD-style brief for a prototype build**
Clone target: `https://spun.global/id` · Mode: **content overhaul (内容爆改)** — keep SPUN's information architecture, rhythm and visual grammar, replace the domain with outbound medical tourism and price estimation.
Status: draft for your review · Written 2026-07-30

---

## 1. Intent summary

SPUN is a **visa** platform: search a corridor → see priced options → upload documents → get an AI-verified outcome, with add-ons and a paid pre-check ("Approval Insight", IDR 99.000).

We are reusing that exact shape for a different, structurally identical problem: an Indonesian patient who has a **resume medis / diagnosis** and needs to know *how much treatment abroad will cost, item by item*.

The mapping is close to 1:1, which is why this clone works:

| SPUN (visa) | SPUN Medika (medical tourism) |
|---|---|
| "Butuh visa ke mana?" — Dari / Ke / Tujuan | "Butuh perawatan apa?" — Dari / Ke / Kondisi atau prosedur |
| Visa card: `Dari IDR 239.000`, `Selesai dalam 4 hari` | Procedure card: `Dari IDR 32.800.000`, `Rawat inap ~1 hari` |
| Approval Insight — upload docs, AI predicts approval | **Estimasi dari Resume Medis** — upload resume medis, AI extracts the case, we itemise the cost |
| Add-ons: translation, hotel, flight, insurance | Travel layer: flight, recovery hotel, companion, interpreter, medical-record translation, insurance |
| 3-step tracking: pay → process → receive | 4-step case: estimate → book → treat → follow-up |
| "400+ tipe visa", 99% approval rate | "Reference prices from N hospitals across 5 destinations" |

**The one screen that has to be excellent** is `rincian-estimasi.html` — the itemized estimate. Everything else supports it.

---

## 2. Users and jobs

**Primary — the patient's family decision-maker** (usually an adult child or spouse, not the patient). Holds the resume medis as a PDF or phone photo. Wants a number they can plan around, and to know what the number does *not* include.
- Job: "I have this diagnosis document. Tell me the realistic total, in rupiah, and what could push it higher."

**Secondary — the patient**, often 55+, comorbid, low tolerance for dense UI. Needs large type, plain Indonesian, no jargon without a gloss.

**Tertiary — the facilitator agent** (SPUN's own ops staff, or a KavaLink-style medical travel facilitator). Wants to open a case, adjust assumptions, and send a shareable estimate to the family. This is the reason the estimate needs editable levers rather than a static PDF.

**Market context (real, sourced — see §10):** 970,000+ Indonesian health travellers went to Malaysia in 2025 (RM 2.2b, +23% YoY); Penang alone saw 527,176 foreign patients (+25.94%). Top specialities: health screening, gastroenterology, oncology, cardiology, endocrinology, orthopaedics. Penang now accepts IDR via QRIS. This is a real, large, underserved corridor — the product premise is sound.

---

## 3. The domain model: how a medical tourism estimate is actually built

This is the research core. A credible estimator has to model the real quoting workflow, not invent a slider.

### 3.1 The real workflow (what we're digitising)

1. Patient sends the **resume medis** (Indonesian discharge summary: diagnosis, ICD-10 code, lab results, imaging, current medication, comorbidities) to 2–3 hospitals or a facilitator. Consultation at this stage is free at most hospitals.
2. The hospital's international patient centre reviews it and assigns a **case type** → an internal procedure/package code.
3. The hospital returns an **itemized cost estimate + treatment plan**. Comparing totals alone is the classic mistake — what differs between hospitals is *what's inside* the total: length of stay, implant brand, follow-up visits, medications.
4. The family then discovers the **non-medical layer**, which runs **30–50% on top** of the procedure cost and is almost never in the hospital quote.

### 3.2 Line-item taxonomy — the four groups

Malaysian private hospitals bill in multiple parts (surgeon's fee, anaesthetist's fee, hospital/facility fee), and Malaysia's MoH mandates price display, so itemisation is genuinely obtainable. Our estimate mirrors that structure:

**Group A — Biaya Medis Inti (core medical)**
- Jasa dokter bedah (surgeon's professional fee)
- Jasa dokter anestesi (anaesthetist's fee) — *frequently omitted from loose quotes*
- Kamar operasi / fasilitas (OT time, equipment, disposables)
- Implan / prostesis / konsumabel — *the other most-common hidden cost; scales with stent or implant count*

**Group B — Pemeriksaan & Rawat Inap (workup and stay)**
- Konsultasi spesialis (pre-op)
- Laboratorium (blood panel, coagulation, cross-match)
- Pencitraan (X-ray / CT / MRI / echo / angiogram)
- Clearance anestesi & jantung
- Kamar rawat inap × malam — **priced by room class, which is the single biggest user-controllable lever**
- Kunjungan harian spesialis, keperawatan, obat inap, IV, dressing, makan — hundreds of ringgit per day on top of the room rate

**Group C — Pasca-Perawatan (post-discharge)**
- Obat pulang — *routinely left off the initial quote; can be several hundred USD*
- Kontrol / follow-up consult
- Fisioterapi / rehabilitasi
- Second opinion or teleconsult back home

**Group D — Non-Medis & Perjalanan (travel layer, +30–50%)**
- Tiket pesawat (patient + companion)
- Hotel pemulihan (nights after discharge)
- Pendamping (companion cost, meals)
- Visa / imigrasi
- Asuransi perjalanan medis
- Penerjemahan rekam medis, interpreter
- Transfer bandara ↔ rumah sakit
- Biaya remitansi / selisih kurs (FX spread — real, and material at IDR 100m+ tickets)

### 3.3 Reference price anchors (2024–2026 published data)

Assume **RM 1 ≈ IDR 3.900** — `TODO: confirm the FX rate at build time and expose it as an editable assumption in the UI.`

| Item | Published RM | ≈ IDR | Note |
|---|---|---|---|
| Katarak (day case) | 8.400 | ~32,8 jt | same-day discharge |
| Ganti sendi lutut total | 33.400 | ~130 jt | avg LOS **4,1 hari** |
| Angiogram koroner | 11.700 | ~45,6 jt | avg **2 hari** |
| Angioplasti + stent | 25.000–50.000 | ~97,5–195 jt | **varies by stent count** |
| Kemoterapi / siklus | 4.000–15.000 | ~15,6–58,5 jt | multiple cycles typical |
| Perawatan stroke | 10.000–30.000 | ~39–117 jt | |
| MRI otak | 1.500–3.500 | ~5,9–13,7 jt | |
| Konsultasi spesialis saraf | 150–400 | ~585 rb–1,56 jt | |
| Kamar: bangsal / malam | 60–120 | ~234–468 rb | |
| Kamar: twin-sharing / malam | 120–250 | ~468–975 rb | |
| Kamar: single / malam | 220–450 | ~858 rb–1,76 jt | |
| Kamar: suite / malam | 500–1.500 | ~1,95–5,85 jt | |

Sourced from Malaysia's 2026 insurance-industry private hospital price guide (2024 data), KPJ/Sentosa's 2025–26 cost guide, and Kavacare's Indonesian-language Malaysia cost pages. **Every number in the build must trace to one of these; nothing gets eyeballed.**

### 3.4 Variance drivers — why the output is a range, never a figure

The estimator's honesty depends on modelling these explicitly. Each is a UI lever or a stated assumption:

| Driver | Effect | Exposed as |
|---|---|---|
| Room class | Up to **25×** between ward and suite, per night | Selector (4 options) |
| Length of stay | Linear on the whole of Group B | Slider, seeded from procedure default (e.g. 4,1 nights for TKR) |
| Implant / stent count | Step function on Group A | Stepper, procedure-conditional |
| Hospital tier / city | Penang runs **10–20% cheaper than KL**, and far cheaper than Singapore | Destination + hospital switcher |
| Comorbidities & stage | Extends workup and LOS; widens the band | Chips read from the resume medis |
| Complications | Tail risk | Explicit "dana kontingensi" row, off by default |
| Malaysian SST on foreign patients | Adds to the bill; under active policy debate | Line item, flagged as policy-dependent |
| FX rate | Whole-total scalar | Editable assumption in the header |
| Companion travel | Doubles most of Group D | Toggle |

### 3.5 Confidence model — non-negotiable

Each line carries a **provenance badge**, and the total's confidence is the weakest link:

- `Harga acuan` — from a published price guide or hospital price list
- `Kutipan rumah sakit` — from an actual quote for this case
- `Estimasi` — derived (e.g. LOS × room rate)
- `Belum diketahui` — needs the hospital; shown as an open item, not silently zeroed

Total renders as **low – mid – high** with a confidence chip (Tinggi / Sedang / Rendah) plus a short "yang bisa menggeser angka ini" list. `Belum diketahui` items are listed *outside* the total, never buried in it.

### 3.6 Safety boundary — the load-bearing rule

The AI reads the resume medis; it does not decide anything clinical.

- Extraction (diagnosis, ICD-10, procedure family, comorbidities, LOS seed) is rendered as an **editable, advisory** review step with the source span quoted from the document and a per-field confidence. The user confirms before any number is computed.
- **Pricing is deterministic.** The model maps text → case fields; a rule table turns case fields → line items. A model output never sets a price directly and never blocks or approves anything.
- Every estimate surface carries: *"Estimasi biaya, bukan penawaran resmi maupun nasihat medis. Rencana perawatan final ditentukan oleh dokter di rumah sakit tujuan."*
- No triage, no severity scoring, no "you should have surgery" language. If the extraction is ambiguous, we say so and ask — we do not guess a diagnosis.

`TODO (you decide): do we want a "kirim ke rumah sakit untuk kutipan resmi" handoff in v1, or does the prototype stop at the estimate?`

---

## 4. Screens

Each screen ships as its own HTML file. `index.html` is a thin launcher/overview linking them.

| # | File | What it is | Priority |
|---|---|---|---|
| 1 | `index.html` | Launcher: screen map with thumbnails | P1 |
| 2 | `beranda.html` | Home — the SPUN home clone, medical content | **P0** |
| 3 | `estimasi-resume-medis.html` | Upload resume medis → AI extraction review (editable) | **P0** |
| 4 | `rincian-estimasi.html` | **The itemized estimate.** The hero screen | **P0** |
| 5 | `prosedur-detail.html` | Procedure/package detail — mirrors the visa detail page | P1 |
| 6 | `bandingkan-rumah-sakit.html` | 3 hospitals side by side, differing inclusions | P1 |
| 7 | `paket-perjalanan.html` | Travel layer / add-ons | P2 |
| 8 | `dasbor-kasus.html` | Case tracking, 4 steps | P2 |

### 4.1 `beranda.html` — section order (traced from the captured original)

Mirrors the original's rhythm exactly; only content changes.

1. **Header** — logo, `Semua Prosedur` / `Untuk Perusahaan` / `Blog`, search icon, `IDR - ID` pill, `Masuk`, `Daftar` (orange).
2. **Hero** — warm sunrise radial gradient, `h1` eyebrow "Platform Berobat ke Luar Negeri" (24px/500) over `h2` "Butuh perawatan apa?" (48px/700), then the pill search bar: **Dari** (kota asal) / **Ke** (negara atau rumah sakit) / **Kondisi atau prosedur** + `Cari`.
3. **Category chip row** — horizontally scrolling pills: Skrining Kesehatan · Jantung · Onkologi · Ortopedi · Gastro · Saraf · Mata · Fertilitas. (Original: Trending Visas, 100% Online, Schengen…)
4. **Filter chip row** — `Filter` · `Estimasi biaya` · `Lama rawat` · `Populer`.
5. **Procedure card grid** (3 cols) — white card, destination image right, title left, `Rawat inap ~4 hari`, `Dari IDR 130.000.000`, strikethrough comparison against Indonesian private-hospital price, savings badge in rose. Cards reference §3.3 anchors only.
6. **Estimasi dari Resume Medis band** — the Approval Insight analogue. Rose-tinted panel, copy left, product screenshot right, CTA `Mulai estimasi`. Free, unlike SPUN's paid pre-check — with a paid `Second opinion` upsell noted beneath.
7. **Layanan pendamping** (3 cards w/ illustration) — Penerjemahan Rekam Medis, Pendampingan & Interpreter, Hotel Pemulihan. `Dari IDR …`.
8. **Cara kerja** — tab pills + product screenshot left, numbered steps right in orange: (1) unggah resume medis, (2) kami itemisasi biaya & konfirmasi ke rumah sakit, (3) berangkat dengan pendampingan, (4) kontrol & rekam medis kembali ke dokter Anda.
9. **Promo** — 3 pale-rose cards with orange tag.
10. **Coverage** — "5 negara, 20+ rumah sakit mitra" as a scattered pill cloud with flags (Malaysia, Singapura, Thailand, India, Korea Selatan), then a stat strip. **Every stat must be real or labelled as illustrative — no invented numbers.**
11. **Testimoni** — masonry review cards. `Original SPUN review text and reviewer names must NOT be copied` — write neutral placeholder testimonials explicitly labelled as sample content.
12. **Footer** — warm cream `#fff6ed`, 4 columns, socials, legal.

### 4.2 `rincian-estimasi.html` — the hero screen, in detail

**Layout:** two columns at ≥1024px. Left (~62%) is the itemized ledger; right is a sticky summary + levers. Single column on mobile with the summary collapsing to a bottom bar.

**Left column**
- Case header: diagnosis + ICD-10 chip, destination, hospital, room class, LOS — each editable inline, each showing whether it came from the document or from a default.
- Four collapsible groups (A/B/C/D from §3.2). Each row: item name · qty × unit · unit range · subtotal range · `Termasuk paket RS?` badge · provenance badge (§3.5).
- **"Sering terlewat"** callout pinned above Group C, flagging the three research-verified omissions: anestesi, implan, obat pulang.
- Open items block: `Belum diketahui` rows, listed outside the total with what's needed to close each.

**Right column (sticky)**
- Total as a range: `IDR 118 jt – 164 jt`, mid emphasised, RM shown secondary.
- Confidence chip + "yang bisa menggeser angka ini" (top 3 variance drivers for this case).
- Live levers: room class (4), LOS slider, stent/implant stepper, companion toggle, contingency toggle, hospital tier switcher, currency IDR/RM, FX rate field. Every change recomputes the ledger live — this is a working prototype, not a screenshot.
- Split donut or stacked bar: medical vs travel share, so the **+30–50% travel layer** is visible rather than a footnote.
- Actions: `Unduh PDF estimasi`, `Kirim via WhatsApp`, `Bandingkan rumah sakit`.
- The disclaimer from §3.6, always visible — not behind a tooltip.

**Acceptance checks**
- Every row's subtotal equals qty × unit range; group totals equal the sum of rows; the header total equals the sum of enabled groups. Numbers must reconcile — a prototype that doesn't add up destroys the whole premise.
- Changing room class from bangsal to suite must visibly move the total (§3.4 says up to 25×/night — if it doesn't, the model is wrong).
- No line item exists without a provenance badge.
- No `Belum diketahui` item is counted in the total.
- The disclaimer is present on every screen that renders a number.

### 4.3 `estimasi-resume-medis.html`

Upload zone (PDF / photo / multiple pages) → "AI membaca dokumen Anda" progress → **extraction review**: a two-pane layout with the document page left and extracted fields right, each field showing the quoted source span, a confidence dot, and an edit affordance. `Konfirmasi & hitung estimasi` is disabled until required fields (diagnosis, procedure family, destination) are confirmed.

Fields extracted: diagnosis (+ICD-10), procedure family, stage/severity where stated, comorbidities, current medication, recent labs/imaging already done (these *subtract* from Group B — a real cost saving worth surfacing), age, allergies.

`TODO: no real medical documents will be used. Build with a synthetic resume medis — fabricated patient, plausible structure. Never a real patient's record.`

---

## 5. Visual system — measured from the original, not guessed

Extracted from `spun.global/id` via computed styles (evidence: `RECON/original-recon-summary.md`, `RECON/screenshots/original-1440.png`, tokens in `design-dna.json`).

| Role | Value | Where it appears |
|---|---|---|
| Canvas | `#fafafa` | body, header, main |
| Surface | `#ffffff` | cards, inputs, chips |
| Warm surface | `#fff6ed` | footer, promo bands |
| Neutral fill | `#efeceb` | inactive chips |
| Ink | `#000000` | headings, body |
| Ink (control) | `#171412` | button labels |
| Muted | `#89766c` | secondary copy |
| Muted light | `#bcb0a9` | tertiary, meta |
| Border | `#dbd5d1` | control outlines |
| Hairline | `#e5e7eb` | dividers |
| **Accent** | `#ff7524` | prices, step numbers, `Daftar`, primary CTA |
| Accent secondary | `#db4268` | discount/savings badges, promo tint |

**Type:** `Instrument Sans` throughout — one family, weights 400/500/600/700. Display `h2` 48px/700, section `h2` 30px/600, `h1` eyebrow 24px/500, card titles 14px/600, body 16px/400, controls 14px/600. `Instrument Sans` is on Google Fonts (OFL) so we self-host the real face — no system-font substitute.

**Geometry:** pill (`9999px`) for chips, inputs, filters, secondary buttons and search; `8px` for header auth buttons and promo CTAs. Cards are large-radius rounded rectangles. Borders are 1px `#dbd5d1`; shadows are very soft and low — border-led containment, not elevation.

**Motion:** Next.js + Tailwind, no Lenis / GSAP / Three detected. Native scroll, horizontal chip scrollers, sticky header. Do not add a smooth-scroll library the original doesn't have.

**Hero treatment:** a warm peach → white radial gradient with a soft sunburst arc behind the search bar. This is the site's one flourish; keep it to the hero only.

> **Design system note.** Your saved default is the Apple-inspired system, but the charter puts a user-provided brand source above it — the clone target *is* the brand source here, so SPUN's measured tokens govern this build. The clinical-safety token set from your verified rules stays in force for any acuity/status signal (§6).

### 5.1 The one place we deviate from SPUN

SPUN has no semantic colour layer. A medical product needs one. We add the clinical-safety tokens already registered in your rules — `#16a34a` / `#d97706` / `#dc2626` — used **only** for estimate-confidence and open-item status, each always paired with an icon and a text label, never as a large fill. Triage/acuity colours are *not* used: this product does not triage.

---

## 6. Content and data model

Seed data lives in a single JS object so the prototype is real and adjustable:

```
destinations[]   → { id, country, city, tierMultiplier, currency, fxToIDR, sstApplies }
hospitals[]      → { id, destinationId, name, tier, specialities[], includedInPackage[] }
procedures[]     → { id, name_id, specialityId, icd10[], losDefault, roomDefault,
                     lineItems[] }   // each with { group, name_id, unit, qtyRule, rmLow, rmHigh, provenance }
roomClasses[]    → { id, name_id, rmPerNightLow, rmPerNightHigh }
travelItems[]    → { id, name_id, idrLow, idrHigh, perPerson, optional }
cases[]          → { id, extracted{}, confirmed{}, levers{} }
```

`TODO: confirm scope — 8 procedures across the 6 top specialities (screening, cardio, onco, ortho, gastro, neuro), 3 destinations (Penang, KL, Singapore) is my recommendation for a prototype that feels complete without inventing data we can't source.`

**Copy language:** Indonesian throughout (the target is `/id`), plain register, no English medical jargon without a gloss. Prices formatted `IDR 130.000.000` with a `Rp 130 jt` short form in dense contexts.

---

## 7. Responsive

- Breakpoints 390 / 768 / 1440, matching the recon captures.
- `rincian-estimasi.html` is the hard one: at 390px the ledger becomes a single column of stacked group cards, the summary becomes a sticky bottom bar showing the mid total plus an expand handle, and the levers move into a bottom sheet. **The itemized table must not horizontally scroll on mobile** — it re-flows into rows.
- Touch targets ≥44px. Older primary users → body never below 16px on mobile.

---

## 8. Explicitly out of scope

Per the clone skill's default boundaries: no login, no payment, no real hospital APIs, no booking confirmation, no PDF generation backend (the download is a print stylesheet), no real patient data, no actual second-opinion service. No clinical decision support of any kind.

---

## 9. Build sequence

1. Complete recon per the web-clone skill: `asset-harvest` for the real Instrument Sans faces, `route-crawl` for the visa-detail template, `interaction-probe` for the chip scrollers and search dropdown. *(Partially done — palette, type, geometry and full-page capture are in `RECON/`.)*
2. Freeze `design-dna.json` (already scaffolded from measured values).
3. Build the price model + seed data first, in isolation. Reconcile the arithmetic before any styling.
4. Build `rincian-estimasi.html` — hero screen first, so the hard problem is solved while there's room to move.
5. Build `beranda.html` against the §4.1 section list.
6. Build `estimasi-resume-medis.html`, then `prosedur-detail.html`, `bandingkan-rumah-sakit.html`.
7. P2 screens, then `index.html` launcher.
8. Generate real imagery for hero and destination cards (no hand-drawn schematics, no stock-photo hot-linking).
9. Write `NOTES.md`, `CLONE_REPORT.md`, `CLONE_AUDIT.md`; run `audit-clone --strict`.

**Pre-deploy replace list (`CLONE_AUDIT.md` will track this):**
- SPUN logo/wordmark — not copied. Set the name in Instrument Sans 600, tracking -0.01em.
- SPUN's testimonial text, reviewer names and Google rating — not copied; sample content, labelled.
- SPUN's product screenshots — not copied; we render our own UI in the mockup frames.
- All tracking (GA / gtag / GTM / pixels) stripped.
- `spun.global` is a live commercial site; this is a visual-language study, not a redeployment.

---

## 10. Open questions

1. **Brand.** I've assumed "SPUN Medika" as a vertical of SPUN, keeping SPUN's DNA. Alternative: a distinct brand that only borrows the layout language. Which?
2. **Fidelity.** Project metadata didn't specify — I've assumed **high-fidelity**. Say the word if you want wireframes first.
3. **Platform.** Assumed **responsive web** (the original is responsive web). Confirm if you want a mobile-first app framing instead.
4. **Destinations.** Assumed Penang / KL / Singapore, with Thailand and India as coverage-only. Adjust?
5. **FX rate.** Assumed RM 1 = IDR 3.900. Confirm or let me pull a live figure at build time.
6. **Depth vs breadth.** 3 excellent screens, or all 8 at moderate depth? My recommendation: 4 P0/P1 screens done properly, P2 as stubs in the launcher.

### Sources

- [Indonesian patients drive Malaysian medical tourism growth — NST](https://www.nst.com.my/amp/news/nation/2026/06/1459086/indonesian-patients-drive-malaysian-medical-tourism-growth)
- [Penang medical tourism revenue crosses RM1.1b — Malay Mail](https://www.malaymail.com/news/malaysia/2026/05/11/penang-sees-surge-in-medical-tourists-revenue-crosses-rm11b-in-2025/219502)
- [New private hospital price list — Malay Mail](https://www.malaymail.com/news/malaysia/2026/02/05/can-you-afford-private-healthcare-new-price-list-helps-malaysians-budget-for-medical-needs/208041)
- [Treatment cost at Malaysian private hospitals, 2025/2026 guide — KPJ Sentosa](https://kpjsentosa.com.my/how-much-does-treatment-cost-at-private-hospitals-in-malaysia-a-complete-2025-2026-guide/)
- [Cost of Medical Treatments in Malaysia for Foreign Patients — Kavacare](https://www.kavacare.id/en/cost-of-medical-treatments-in-malaysia/)
- [Berobat ke Malaysia dan Penang: langkah dan estimasi biaya — Kavacare](https://www.kavacare.id/berobat-ke-malaysia-dan-penang-langkah-dan-estimasi-biaya/)
- [Biaya pengobatan saraf di Malaysia — Expediheal](https://expediheal.com/news-blog/biaya-pengobatan-saraf-di-malaysia-estimasi-dan-faktor-yang-mempengaruhi)
- [Medical tourism checklist — Sylk Health](https://sylkhealth.com/blog/articles/medical-tourism-checklist)
- [Comparing medical tourism prices — TravellerMD](https://www.travellermd.com/post/how-do-you-compare-prices-across-different-medical-tourism-destinations)
- [Impact of increased SST on medical tourists in Malaysia — Penang Institute](https://penanginstitute.org/wp-content/uploads/2025/10/Assessing-the-Impact-of-Increased-SST-on-Medical-Tourists-in-Malaysia-A-Focus-on-Penang.pdf)
- [Medical tourism vs US healthcare cost comparison](https://www.medicaltourismpackages.com/medical-tourism-vs-us-healthcare-cost-comparison/)

---

## Next step

**Read this document and edit it directly** — especially §4 (screen list), §6 (`TODO` on scope) and §10 (the six open questions). Change numbers, cut screens, rewrite the Indonesian copy direction; the file is the spec, so whatever you leave here is what gets built.

When it reads right, reply **"build from plan.md"** (add answers to any of §10 you care about) and I'll switch to Design mode and start at step 3 of §9 — price model first, then `rincian-estimasi.html`.