Langhui
AI+ Biopharmaceutical Data Infrastructure

医薬品データ
Drive the A.I. Pharmaceutical future

In the area of deep-scale AI+ biopharmaceuticals, Longhui Technologies has built 18 high-quality data sets covering the entire life cycle of drug development, ranging from target detection, compound screening, ADMT prediction to clinical trials, drug repositioning, and providing full-link data support for the AI pharmaceutical model from early detection to post-market evaluation.

18
医薬品データセット
200万人以上
化合物分子
500,000 +
ターゲットフォーマットペア
100,000+以上
Clinical trial records
ai-drug-discovery.py
import langhuiai as lh

# Loading the compound reservoir and molecular characteristics
compounds = lh.pharma.load_compounds(
  source='ZINC/ChEMBL/PubChem',
  count=200000000
)

# Loading target-assemble combination data
target_ligand = lh.pharma.load_target_ligand(
  targets=500000,
  assays='IC50/Ki/Kd'
)

# ADMET Training Series on Forecasting
admet = lh.pharma.load_admet(
  endpoints=50,
  molecules=1000000
)

# Clinical trial data
trials = lh.pharma.load_clinical_trials(
  phases='I/II/III/IV'
)

print("The Ai pharmaceutical data is loaded.")
Advances in the AI Pharmaceutical Front

2025-2026 Biopharmaceutical AI breakthrough progress

From AlphaFold 3 to AI Design Medicine Clinic III, from Protein Design to Drug Repositioning, AI is remodeling the drug paradigm

Nature 2025

AlphaFold 3 Protein Combined Forecast

DeepMind publishes AlphaFold 3, which achieves the first prediction of a protein + RNA+DNA+ small molecular composition combined structure, far more accurate than AlphaFold 2. Nature 2025 reviews this as a "new era of structural biology" and reduces the structure of drug targets from months to minutes.

Structure predicts a thousand times faster.
Science 2025

AI design drugs entering clinical phase III

Insilico Medicine's AI design special pulmonary fibrosis drug, INS 018_055, entered clinical phase IIa and displayed positive data, becoming the first global AI candidate for design and clinical entry. Science commented that AI pharmaceuticals moved from conceptual validation to clinical landing.

50% reduction in R & D cycle
Cell 2025

Generating AI design from head protein

Models such as RFdiffusion, ProteinMPNN, ESM3 have been designed to design a new protein that does not exist in nature. The study published by Cell 2025 demonstrates that AI designed to combine protein proton protege and force beyond natural antibodies, and to open a new paradigm for biopharmaceutical development.

Design success rate 90%+
NEJM 2025

Repositioning of AI-driven drugs

AI analyses the real world data and molecular characteristics of big amounts, providing clues for new uses of old drugs. A study published by NEJM 2025 shows that AI repositioned to identify Baretini for the treatment of COVID-19 serious diseases, from hypothetical to clinical validation only six months, reflecting the clinical value of AI repositioning.

Reduction in the certification cycle 70 per cent
Nat Biotech 2026

The target has a large model.

The multi-group megamodels (scGPT, Geneformer, BioGPT) identify new target points from single-cell sequencing, gene expression spectrum. NAture Biotechnology 2026 studies show that new targets identified by AI show potential for difficult diseases such as Alzheimer ' s disease, non-alcoholic fatty liver, etc.

The new target has been found three times higher.
Lancet 2026

AI drug safety forecast

The in-depth learning-based ADMT prediction model has an accurate rate of overestimation of side effects such as hepatotoxicity, heart toxicity (HERG), kidney toxicity, etc. The Lancet 2026 paper stressed that AID drug safety predictions can reduce clinical trial failure rates by about 30%.

We're down to 30%.
Langhuiのレイアウト

Long Salang Hui Medical Data Strategy Layout

Build a high-quality data infrastructure that covers the life cycle of drug research and development and empowers AI pharmaceutical innovation

創薬

200 million + compound molecular structure, bioactive data to support virtual screening and molecular generation

4つのデータセット

ADMET予測

50+ADMT endpoint, 1 million+molecule experimental data, predicting drug metabolic and toxicity

3個のデータセット

臨床試験

100,000+Data for clinical trials covering full I-IV process, including entry/end-of-service/bad events

4つのデータセット

薬剤の新規位置づけ

Drug-disease linkages based on real world data, supporting new excavations of old drugs

3個のデータセット

Pharmacist+Scientific Institutes Network

Data cooperation mechanisms have been established with the top scientific institutes of 30+ innovative drug companies in China, CRO and Shanghai Institute of Medicine at the Chinese Academy of Medicine, and the Chinese University of Pharmacy to ensure the authority, coverage and professional designation of data sources.

Cooperation coverage 88 per cent30+ Cooperative Agency

Pharmacological expert identification + auditing system

Doctorate in Pharmacy, expert examination in Pharmacology/Pharmacology, AAA Auxiliary Quality Control System, Level 3 system to ensure the professional and accurateness of target-formulation data, ADMT labels, clinical end-of-doctoral interpretation.

アノテーション精度 99.5%80+ファーマリストスペシャリスト

Full-link data compliance assurance

Strictly following NMPA, FDA, EMA Data Management Guidelines for Drug Research and Development and GCP (Merticultural Quality Management Guidelines for Drug Clinical Trials), all clinical data are dissensitized and have a complete authorized link.

NMPAコンプライアンス FDAの標準 GCPコード

AI+ Pharmacological Synergetic Syndicate Platform

The self-study pharmacological dedicated pedestal platform supports molecular structure visualization, SMILES automatic verification, target-formulation mapping, end-of-clinical trial interpretation aids, and human synergy increases the marking efficiency by three to five times.

SMILES standard PDB形式 Chembl compatible
データセットマトリクス

18 medical data sets at a glance

By R & D stage, covering the four main sections of drug discovery, ADMT, clinical trials, drug repositioning

Compound Molecular Bank

ディスカバリー

200 million + compound molecular structure (SMILES/InChI) with ZINC, ChEMBL, PubChem Multisource Integration

200 million+molecules 詳細 →

Target-format combination data

靶点

500,000 + target-format pairs, marked with IC50-Ki/Kd experimental values and integration points

500,000+Choose 詳細 →

Protein Structure Database

靶点

200,000+Protein3D Structure (PDB format), containing AlphaFold Forecast and Experimental Parsing Structure

200,000+Structure 詳細 →

Bioactive data sets

ディスカバリー

15 million + compound-target bioactivity records covering various types of action such as inhibition/activation/coupling

15 million + records 詳細 →

ADMET予測データセット

ADMET

50+ADMT endpoint, 1 million+molecule experimental data, covering solubility/permeability/resiliency

1 million + molecules 詳細 →

Dataset on drug toxicity

毒性

Multiple endpoint toxicity data, such as hepatotoxicity/cardiology (HERG)/renal toxicity/reproductive toxicity

500,000+サンプル 詳細 →

Data on drug metabolic

ADMET

CYP450 E.S. metabolic stability, metabolic product identification, drug-drug interaction data

300,000+Records の 詳細 →

Registration data for clinical trials

臨床検査

100,000+registration of clinical trials covering I-IV period, including entry/exclusion criteria/endpoint indicators

100,000+test の 詳細 →

Data on clinical trials

臨床検査

50,000 + report on clinical trial results, including main endpoint/secondary endpoint/bad event statistics

50,000+ レポート 詳細 →

リアルワールド・エイビデンス(RWE)

臨床検査

2 million + patient real world drug data to support post-market evaluation and precision medical care

2 million+ patients 詳細 →

Data for bad incident reporting

安全

FDA FAERS database structurally deficient incident report, 5 million plus bad incident record

5 million + report 詳細 →

Data on drug-disease correlation

重位置づけ

A network of drug-disease links based on literature mining and real world evidence to support new uses of old drugs

800,000+コンテンツ 詳細 →

ドラッグターゲットネットワーク

重位置づけ

Multi-target drug network, predicting drug detargeting and repositioning opportunities

1 million+relation 詳細 →

Listed drug database

重位置づけ

Global complete information on listed drugs, including adaptive/role mechanisms/agent types/paths

15 000 + Drugs 詳細 →

Text of the drug instruction book

ディスカバリー

50,000+ Structured text of drug instructions with a degree of adaptability/use/nox/negative reaction

50,000+ instructions 詳細 →

Drug Interaction (DI)

安全

2 million + drug-drug interactions, mechanism of effect/severity/clinical advice

Two million plus DDI. 詳細 →

バイオマークデータ

臨床検査

Drugs respond to biomarker data, supporting precision drug use and accompanying diagnostic development

100,000+ロゴ 詳細 →

Natural product dataset

ディスカバリー

300,000 + molecular structure and biological activity of natural products supporting modernization of Chinese medicine and discovery of new drugs

300,000+molecular 詳細 →
データサンプル

Presentation of medical data labeling format

View the label structure and illustrative data of the different types of pharmaceutical data set

pharma-sample.json
{
  "compound_id": "ZINC000000001",
  "name": 「アスピリン」,
  "smiles": "CC(=O)OC1=CC=CC=C1C(=O)O",
  "molecular_formula": "C9H8O4",
  "molecular_weight": 180.16,
  "logP": 1.19,
  "targets": [
    {
      "target": "COX-1",
      "ic50": 0.0015,
      "unit": "μM"
    }
  ],
  "indication": "Salvation/anti-inflammation/anti-blood plate."
}
活用シーン

AI+ Biopharmaceutical application scene

ロングハイ・メディス・データ・セットがサポートするコア・アプリケーション・シーン

Target Discovery and Validation

Based on multi-group data and knowledge maps, AI identifies new targets and validates the drug properties

Virtual filter and molecular generation

Screening of candidate molecules from the billion-grade compound reservoir or generation of a completely new molecular structure from scratch

ADMET予測

Early prediction of drug metabolism and toxicity, reducing risk of post-clinical failure

薬剤の新規位置づけ

Digging new adaptations for already marketed drugs, reducing the research and development cycle and costs

Clinical trials optimized

AI-assisted patient entry screening, end-point prediction, trial design optimization

Drug safety monitoring

A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A. A.

よくあるご質問

Common problems with pharmaceutical data

A common question answer for the Long Hye Medicine Data Set

How do pharmaceutical data ensure compliance?

Langhui strictly follows the data management requirements of regulatory bodies such as NMPA (National Drug Authority), FDA and EMA, and the GCP (Quality Management Guidelines for Drug Clinical Trials).

How do target-format combination data ensure accuracy?

Target-format data are derived from authoritative public databases such as ChEMBL, BindingDB, PDFbind and internal experimental data in collaboration with pharmaceutical companies.

Which molecular formats are supported by the data set?

Supports various standard molecular formats: SMILES (linear molecular expression), InCHI/InchiKey (international chemical marking), MOL/SDF (molecule structure document), PDB (protein 3D structure), CIF (crystalcology information file).

Can drug data be customized for specific disease areas?

Langhui provides customized data services that collect, integrate, and label drug data in specific disease areas according to the client’s research orientation (e.g. tumors, neurodegenerative diseases, metabolic diseases, etc.). We have a network of 30+ companies working with research institutes that respond quickly to customized needs, with a normal 6-10 week cycle from data collection to the delivery of markers.

How can medical datasets be accessed and used authorized?

We will provide the corresponding data authorization program based on your use scene (academic research/business research/development/education training) and sign the Data Use Agreement (DUA) to ensure data use compliance and intellectual property protection.

Building the data base of the AI Pharmaceuticals

Langhui Medicine Dataset, providing high-quality, compliant and trusted data support for AI+ biopharmaceutical innovation