diff --git a/docs/index.rst b/docs/index.rst index 3554861..2784f56 100644 --- a/docs/index.rst +++ b/docs/index.rst @@ -21,6 +21,7 @@ Welcome to TorchSurv's documentation! notebooks/introduction notebooks/momentum notebooks/regression_time_varying + notebooks/synthetic_data_signal notebooks/non_medical_applications .. toctree:: diff --git a/docs/notebooks/synthetic_data_signal.ipynb b/docs/notebooks/synthetic_data_signal.ipynb new file mode 100644 index 0000000..6dc9fe3 --- /dev/null +++ b/docs/notebooks/synthetic_data_signal.ipynb @@ -0,0 +1,273 @@ +{ + "cells": [ + { + "cell_type": "markdown", + "id": "0f57957c", + "metadata": {}, + "source": [ + "# Synthetic Survival Data With Controlled Signal\n", + "\n", + "This notebook demonstrates how to use `torchsurv.tools.make_synthetic_data` to create Cox-style synthetic survival datasets with controllable signal strength.\n", + "\n", + "The goal is to show a practical failure mode from [issue #154](https://github.com/Novartis/torchsurv/issues/154): when the loss does not improve, the problem can come from the data having weak or no learnable survival signal, not from `TorchSurv` itself.\n", + "\n", + "Here `rho` controls how much of the true latent risk is explained by the covariates:\n", + "\n", + "- `rho = 1.0`: very easy, features fully determine the latent risk\n", + "- `rho = 0.0`: very hard, features contain no risk information\n", + "- intermediate values: gradually harder learning problems\n" + ] + }, + { + "cell_type": "code", + "execution_count": 1, + "id": "29f8f2ff", + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/Users/corolth1/anaconda3/envs/torchsurv/lib/python3.10/site-packages/torchvision/io/image.py:13: UserWarning: Failed to load image Python extension: 'Could not load this library: /Users/corolth1/anaconda3/envs/torchsurv/lib/python3.10/site-packages/torchvision/image.so'If you don't plan on using image functionality from `torchvision.io`, you can ignore this warning. Otherwise, there might be something wrong with your environment. Did you have `libjpeg` or `libpng` installed before building `torchvision` from source?\n", + " warn(\n" + ] + }, + { + "data": { + "text/plain": [ + "" + ] + }, + "execution_count": 1, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "import matplotlib.pyplot as plt\n", + "import torch\n", + "from torch import nn\n", + "\n", + "from torchsurv.loss.cox import neg_partial_log_likelihood\n", + "from torchsurv.metrics.cindex import ConcordanceIndex\n", + "from torchsurv.tools import make_synthetic_data\n", + "\n", + "torch.manual_seed(7)" + ] + }, + { + "cell_type": "code", + "execution_count": 2, + "id": "626364a3", + "metadata": {}, + "outputs": [], + "source": [ + "def split_batch(batch, train_fraction: float = 0.8):\n", + " n = batch[\"x\"].shape[0]\n", + " n_train = int(train_fraction * n)\n", + " permutation = torch.randperm(n)\n", + " train_index = permutation[:n_train]\n", + " val_index = permutation[n_train:]\n", + "\n", + " train_batch = {\n", + " key: value[train_index] if value.ndim > 0 and value.shape[0] == n else value for key, value in batch.items()\n", + " }\n", + " val_batch = {\n", + " key: value[val_index] if value.ndim > 0 and value.shape[0] == n else value for key, value in batch.items()\n", + " }\n", + " return train_batch, val_batch\n", + "\n", + "\n", + "def train_cox_model(train_batch, val_batch, *, epochs: int = 120, lr: float = 0.05):\n", + " model = nn.Linear(train_batch[\"x\"].shape[1], 1)\n", + " optimizer = torch.optim.Adam(model.parameters(), lr=lr)\n", + " cindex = ConcordanceIndex()\n", + "\n", + " train_losses = []\n", + " val_cindices = []\n", + "\n", + " for _ in range(epochs):\n", + " optimizer.zero_grad()\n", + " loss = neg_partial_log_likelihood(model(train_batch[\"x\"]), train_batch[\"event\"], train_batch[\"time\"])\n", + " loss.backward()\n", + " optimizer.step()\n", + " train_losses.append(loss.detach())\n", + "\n", + " with torch.no_grad():\n", + " risk_val = model(val_batch[\"x\"]).squeeze(-1)\n", + " val_cindices.append(cindex(risk_val, val_batch[\"event\"], val_batch[\"time\"], instate=False).detach())\n", + "\n", + " return model, torch.stack(train_losses), torch.stack(val_cindices)" + ] + }, + { + "cell_type": "code", + "execution_count": 3, + "id": "7c89d941", + "metadata": {}, + "outputs": [], + "source": [ + "def run_experiment(rho: float, *, n: int = 512, m: int = 8, seed: int = 7):\n", + " batch = make_synthetic_data(n=n, m=m, rho=rho, censoring_rate=0.3, seed=seed)\n", + " train_batch, val_batch = split_batch(batch)\n", + " _, train_losses, val_cindices = train_cox_model(train_batch, val_batch)\n", + " return {\n", + " \"rho\": rho,\n", + " \"train_losses\": train_losses,\n", + " \"val_cindices\": val_cindices,\n", + " \"censoring_rate\": (~batch[\"event\"]).float().mean(),\n", + " }" + ] + }, + { + "cell_type": "code", + "execution_count": 4, + "id": "f472bc1a", + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "{1.0: {'final_train_loss': 4.810527324676514,\n", + " 'final_val_cindex': 0.7056451439857483,\n", + " 'observed_censoring_rate': 0.298828125},\n", + " 0.75: {'final_train_loss': 4.89111328125,\n", + " 'final_val_cindex': 0.7180487513542175,\n", + " 'observed_censoring_rate': 0.298828125},\n", + " 0.5: {'final_train_loss': 4.932214260101318,\n", + " 'final_val_cindex': 0.6513043642044067,\n", + " 'observed_censoring_rate': 0.298828125},\n", + " 0.25: {'final_train_loss': 5.0428667068481445,\n", + " 'final_val_cindex': 0.6064745783805847,\n", + " 'observed_censoring_rate': 0.298828125},\n", + " 0.0: {'final_train_loss': 5.140138149261475,\n", + " 'final_val_cindex': 0.4678764045238495,\n", + " 'observed_censoring_rate': 0.298828125}}" + ] + }, + "execution_count": 4, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "rhos = [1.0, 0.75, 0.5, 0.25, 0.0]\n", + "results = [run_experiment(rho) for rho in rhos]\n", + "\n", + "summary = {\n", + " result[\"rho\"]: {\n", + " \"final_train_loss\": float(result[\"train_losses\"][-1]),\n", + " \"final_val_cindex\": float(result[\"val_cindices\"][-1]),\n", + " \"observed_censoring_rate\": float(result[\"censoring_rate\"]),\n", + " }\n", + " for result in results\n", + "}\n", + "summary" + ] + }, + { + "cell_type": "code", + "execution_count": 5, + "id": "3ad65b34", + "metadata": {}, + "outputs": [ + { + "data": { + "image/png": "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", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "fig, axes = plt.subplots(1, 2, figsize=(12, 4))\n", + "\n", + "for result in results:\n", + " normalized_loss = result[\"train_losses\"] / result[\"train_losses\"][0]\n", + " axes[0].plot(normalized_loss.numpy(), label=f\"rho={result['rho']:.2f}\")\n", + " axes[1].plot(result[\"val_cindices\"].numpy(), label=f\"rho={result['rho']:.2f}\")\n", + "\n", + "axes[0].set_title(\"Normalized Training Loss\")\n", + "axes[0].set_xlabel(\"Epoch\")\n", + "axes[0].set_ylabel(\"Loss / initial loss\")\n", + "axes[0].set_yscale(\"log\")\n", + "\n", + "axes[1].set_title(\"Validation C-index\")\n", + "axes[1].set_xlabel(\"Epoch\")\n", + "axes[1].set_ylabel(\"C-index\")\n", + "axes[1].axhline(0.5, color=\"black\", linestyle=\"--\", linewidth=1)\n", + "\n", + "for axis in axes:\n", + " axis.legend()\n", + "\n", + "plt.tight_layout()" + ] + }, + { + "cell_type": "code", + "execution_count": 6, + "id": "cd164f9b", + "metadata": {}, + "outputs": [ + { + "data": { + "image/png": "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", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "fig, ax = plt.subplots(figsize=(6, 4))\n", + "ax.plot(rhos, [summary[rho][\"final_val_cindex\"] for rho in rhos], marker=\"o\")\n", + "ax.axhline(0.5, color=\"black\", linestyle=\"--\", linewidth=1)\n", + "ax.set_xlabel(\"rho\")\n", + "ax.set_ylabel(\"Final validation C-index\")\n", + "ax.set_title(\"Performance degrades as signal decreases\")\n", + "plt.tight_layout()" + ] + }, + { + "cell_type": "markdown", + "id": "9fa72d16", + "metadata": {}, + "source": [ + "## Interpretation\n", + "\n", + "This benchmark is meant to separate two cases:\n", + "\n", + "1. When `rho` is high, a simple linear Cox model should reduce the loss and reach a validation C-index clearly above random.\n", + "2. When `rho` is near zero, the same training loop, loss function, and optimization setup can plateau near random ranking because the covariates no longer contain usable signal.\n", + "\n", + "That is the practical point of the synthetic benchmark: if a real experiment does not improve, it may reflect a weak data-generating signal rather than a bug in `TorchSurv`.\n" + ] + } + ], + "metadata": { + "kernelspec": { + "display_name": "torchsurv", + "language": "python", + "name": "python3" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.10.18" + } + }, + "nbformat": 4, + "nbformat_minor": 5 +} diff --git a/pyproject.toml b/pyproject.toml index 943085b..58435dc 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -125,7 +125,7 @@ IssueTracker = "https://github.com/Novartis/torchsurv/issues" Changelog = "https://opensource.nibr.com/torchsurv/CHANGELOG.html" [tool.codespell] -ignore-words-list = ["TE", "FPR", "tOI", "te", "FO", "MIs", "fO", "nd"] # Known false positives +ignore-words-list = ["TE", "FPR", "tOI", "te", "FO", "MIs", "fO", "nd", "ue"] # Known false positives skip = [ "*.bib", "*.toml", diff --git a/src/torchsurv/tools/__init__.py b/src/torchsurv/tools/__init__.py index 974ef29..43b4c0b 100644 --- a/src/torchsurv/tools/__init__.py +++ b/src/torchsurv/tools/__init__.py @@ -1,7 +1,8 @@ -"""This module provides validation utilities for survival analysis inputs.""" +"""Utilities for survival analysis inputs and synthetic benchmarking data.""" from __future__ import annotations +from torchsurv.tools.synthetic import make_synthetic_data from torchsurv.tools.validators import ( EvalTimeInputs, ModelInputs, @@ -18,6 +19,7 @@ "NewTimeInputs", "SurvivalInputs", "TimeVaryingCoxInputs", + "make_synthetic_data", "impute_missing_log_shape", "validate_time_varying_log_hz", ] diff --git a/src/torchsurv/tools/synthetic.py b/src/torchsurv/tools/synthetic.py new file mode 100644 index 0000000..480946e --- /dev/null +++ b/src/torchsurv/tools/synthetic.py @@ -0,0 +1,152 @@ +from __future__ import annotations + +import math + +import torch + +__all__ = ["make_synthetic_data"] + + +def _validate_positive_int(name: str, value: int) -> None: + if not isinstance(value, int): + raise ValueError(f"Input '{name}' must be an integer.") + if value <= 0: + raise ValueError(f"Input '{name}' must be strictly positive.") + + +def _validate_probability(name: str, value: float, *, allow_one: bool) -> float: + if not isinstance(value, (int, float)): + raise ValueError(f"Input '{name}' must be a number.") + value = float(value) + upper_bound = 1.0 if allow_one else 1.0 - 1e-12 + if value < 0.0 or value > upper_bound: + comparator = "[0, 1]" if allow_one else "[0, 1)" + raise ValueError(f"Input '{name}' must be in {comparator}.") + return value + + +def _standardize(x: torch.Tensor) -> torch.Tensor: + return (x - x.mean()) / x.std().clamp_min(torch.finfo(x.dtype).eps) + + +def _calibrate_censoring( + event_time: torch.Tensor, raw_censor_time: torch.Tensor, censoring_rate: float +) -> torch.Tensor: + """Scale raw censoring times to approximately match the requested censoring rate.""" + if censoring_rate == 0.0: + return torch.full_like(event_time, torch.inf) + + lower = torch.tensor(0.0, dtype=event_time.dtype, device=event_time.device) + upper = torch.tensor(1.0, dtype=event_time.dtype, device=event_time.device) + + def observed_censoring(scale: torch.Tensor) -> torch.Tensor: + return (scale * raw_censor_time < event_time).float().mean() + + while observed_censoring(upper) > censoring_rate: + upper = upper * 2.0 + + for _ in range(60): + midpoint = (lower + upper) / 2.0 + if observed_censoring(midpoint) > censoring_rate: + lower = midpoint + else: + upper = midpoint + + return upper * raw_censor_time + + +def make_synthetic_data( + n: int, + m: int, + rho: float, + *, + censoring_rate: float = 0.3, + seed: int | None = None, +) -> dict[str, torch.Tensor]: + """Generate a synthetic survival dataset for Cox-model benchmarking. + + The generated data follow a proportional-hazards construction: + + - features are IID Gaussian, + - the true log-risk is a mixture of feature-derived signal and independent noise, + - event times follow an exponential baseline hazard scaled by ``exp(log_risk)``, + - censoring is independent and calibrated to an approximate target rate. + + Args: + n: Number of samples. + m: Number of features. + rho: Signal strength in ``[0, 1]``. ``rho=1`` means the latent log-risk + is fully determined by the features, while ``rho=0`` means the + latent log-risk is independent of the features. + censoring_rate: Approximate fraction of censored observations in ``[0, 1)``. + seed: Optional random seed for reproducibility. + + Returns: + Dictionary containing: + + - ``x``: Covariate matrix with shape ``(n, m)`` + - ``event``: Event indicator with shape ``(n,)`` and dtype ``bool`` + - ``time``: Observed event/censoring time with shape ``(n,)`` + - ``log_risk``: Ground-truth latent Cox log-risk with shape ``(n,)`` + - ``beta``: Ground-truth normalized feature coefficients with shape ``(m,)`` + + Examples: + >>> batch = make_synthetic_data(n=64, m=8, rho=0.75, seed=7) + >>> sorted(batch.keys()) + ['beta', 'event', 'log_risk', 'time', 'x'] + >>> batch["x"].shape + torch.Size([64, 8]) + """ + _validate_positive_int("n", n) + _validate_positive_int("m", m) + rho = _validate_probability("rho", rho, allow_one=True) + censoring_rate = _validate_probability("censoring_rate", censoring_rate, allow_one=False) + + generator = torch.Generator() + if seed is not None: + generator.manual_seed(seed) + + x = torch.randn((n, m), generator=generator, dtype=torch.float32) + + beta = torch.randn((m,), generator=generator, dtype=torch.float32) + beta = beta / beta.norm().clamp_min(torch.finfo(beta.dtype).eps) + + signal = _standardize(x @ beta) + noise = _standardize(torch.randn((n,), generator=generator, dtype=torch.float32)) + + log_risk = math.sqrt(rho) * signal + math.sqrt(1.0 - rho) * noise + + baseline_hazard = torch.tensor(0.1, dtype=torch.float32) + uniforms = torch.rand((n,), generator=generator, dtype=torch.float32).clamp_min(torch.finfo(torch.float32).eps) + event_time = -torch.log(uniforms) / (baseline_hazard * torch.exp(log_risk)) + + censor_uniforms = torch.rand((n,), generator=generator, dtype=torch.float32).clamp_min( + torch.finfo(torch.float32).eps + ) + raw_censor_time = -torch.log(censor_uniforms) + censor_time = _calibrate_censoring(event_time, raw_censor_time, censoring_rate) + + event = event_time <= censor_time + time = torch.minimum(event_time, censor_time) + + if not event.any(): + first_event = torch.argmin(event_time) + event[first_event] = True + time[first_event] = event_time[first_event] + + return { + "x": x, + "event": event, + "time": time, + "log_risk": log_risk, + "beta": beta, + } + + +if __name__ == "__main__": + import doctest + + # Run doctest + results = doctest.testmod() + if results.failed == 0: + print("All tests passed.") diff --git a/tests/test_synthetic.py b/tests/test_synthetic.py new file mode 100644 index 0000000..4356c11 --- /dev/null +++ b/tests/test_synthetic.py @@ -0,0 +1,91 @@ +import pytest +import torch +from torch import nn + +from torchsurv.loss.cox import neg_partial_log_likelihood +from torchsurv.metrics.cindex import ConcordanceIndex +from torchsurv.tools import make_synthetic_data + + +class TestSyntheticData: + def test_shapes_and_dtypes(self): + batch = make_synthetic_data(n=128, m=6, rho=0.75, seed=12) + + assert set(batch) == {"x", "event", "time", "log_risk", "beta"} + assert batch["x"].shape == (128, 6) + assert batch["event"].shape == (128,) + assert batch["time"].shape == (128,) + assert batch["log_risk"].shape == (128,) + assert batch["beta"].shape == (6,) + assert batch["x"].dtype == torch.float32 + assert batch["event"].dtype == torch.bool + assert batch["time"].dtype == torch.float32 + assert batch["log_risk"].dtype == torch.float32 + assert batch["beta"].dtype == torch.float32 + assert torch.all(batch["time"] >= 0.0) + assert batch["event"].any() + + def test_reproducible_with_seed(self): + batch_a = make_synthetic_data(n=64, m=4, rho=0.5, censoring_rate=0.2, seed=7) + batch_b = make_synthetic_data(n=64, m=4, rho=0.5, censoring_rate=0.2, seed=7) + + for key in batch_a: + assert torch.equal(batch_a[key], batch_b[key]) + + @pytest.mark.parametrize( + ("kwargs", "message"), + [ + ({"n": 0, "m": 4, "rho": 0.5}, "n"), + ({"n": 32, "m": 0, "rho": 0.5}, "m"), + ({"n": 32, "m": 4, "rho": -0.1}, "rho"), + ({"n": 32, "m": 4, "rho": 1.1}, "rho"), + ({"n": 32, "m": 4, "rho": 0.5, "censoring_rate": -0.1}, "censoring_rate"), + ({"n": 32, "m": 4, "rho": 0.5, "censoring_rate": 1.0}, "censoring_rate"), + ], + ) + def test_invalid_inputs_raise(self, kwargs, message): + with pytest.raises(ValueError, match=message): + make_synthetic_data(**kwargs) + + def test_signal_strength_controls_concordance(self): + cindex = ConcordanceIndex() + high_signal = [] + low_signal = [] + + for seed in range(5): + high_batch = make_synthetic_data(n=256, m=8, rho=1.0, seed=seed) + low_batch = make_synthetic_data(n=256, m=8, rho=0.0, seed=seed) + high_estimate = high_batch["x"] @ high_batch["beta"] + low_estimate = low_batch["x"] @ low_batch["beta"] + + high_signal.append(cindex(high_estimate, high_batch["event"], high_batch["time"], instate=False).item()) + low_signal.append(cindex(low_estimate, low_batch["event"], low_batch["time"], instate=False).item()) + + high_mean = sum(high_signal) / len(high_signal) + low_mean = sum(low_signal) / len(low_signal) + + assert high_mean > low_mean + 0.2 + assert 0.4 < low_mean < 0.6 + + def test_high_signal_dataset_is_trainable(self): + batch = make_synthetic_data(n=256, m=8, rho=1.0, seed=0) + model = nn.Linear(8, 1) + optimizer = torch.optim.Adam(model.parameters(), lr=0.05) + cindex = ConcordanceIndex() + + losses = [] + with torch.no_grad(): + initial_cindex = cindex(model(batch["x"]).squeeze(), batch["event"], batch["time"], instate=False).item() + + for _ in range(80): + optimizer.zero_grad() + loss = neg_partial_log_likelihood(model(batch["x"]), batch["event"], batch["time"]) + loss.backward() + optimizer.step() + losses.append(loss.item()) + + with torch.no_grad(): + final_cindex = cindex(model(batch["x"]).squeeze(), batch["event"], batch["time"], instate=False).item() + + assert losses[-1] < losses[0] + assert final_cindex > initial_cindex + 0.1