pith. sign in

arxiv: 2606.26901 · v1 · pith:M7KYNN4Mnew · submitted 2026-06-25 · 💻 cs.CL · cs.AI

SamaVaani: Auditing and Debiasing Multilingual Clinical ASR for Indian Languages

Subham Kumar , Prakrithi Shivaprakash , Abhishek Manoharan , Astut Kurariya , Diptadhi Mukherjee , Prabhat Chand , Pratima Murthy , Koustav Rudra
show 2 more authors
Lekhansh Shukla Animesh Mukherjee
This is my paper

Pith reviewed 2026-06-26 04:45 UTC · model grok-4.3

classification 💻 cs.CL cs.AI
keywords automatic speech recognitiondebiasingmultilingual ASRclinical applicationsfairnessIndian languagespsychiatric interviewsfine-tuning
0
0 comments X

The pith

SamaVaani is a unified debiasing technique that simultaneously improves ASR performance and fairness across demographic groups in clinical interviews.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper first audits eight ASR models on real psychiatric interview recordings in Kannada, Hindi, and Indian English and documents large performance differences by language plus systematic gaps by speaker gender and role. It then fine-tunes two open-source models and presents SamaVaani as a single fairness-aware fine-tuning procedure. The procedure is shown to raise overall recognition accuracy while shrinking the demographic gaps. A reader would care because clinical documentation tools must work reliably and equitably for diverse Indian speakers if they are to be used in healthcare.

Core claim

After auditing models including IndicWhisper, WhisperLargeV3, Sarvam, GoogleS2T, Gemma3n, OmniLingual, Vaani, and Gemini on real psychiatric interview data, the authors fine-tune Gemma3n and OmniLingual and introduce SamaVaani, a unified debiasing technique that simultaneously improves ASR performance and improves fairness across demographic groups.

What carries the argument

SamaVaani, the unified debiasing technique implemented through fairness-aware fine-tuning of ASR models.

If this is right

  • Clinical ASR systems can be made more accurate for Kannada, Hindi, and Indian English by the same fine-tuning step that improves fairness.
  • Gaps tied to speaker role and gender can be narrowed without sacrificing overall word error rate performance.
  • Open-source models can be adapted to reduce the fairness shortfalls observed in both open and commercial systems.
  • Equitable deployment of ASR for documenting psychiatric encounters becomes feasible across the three languages.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same audit-plus-fine-tuning pattern could be applied to other medical speech tasks or additional Indian languages.
  • Reduced demographic gaps might increase clinician and patient willingness to rely on ASR-generated notes.
  • The technique might be tested for side effects on transcription of rare medical terms or code-switching speech.

Load-bearing premise

The performance gaps found in the audit are caused mainly by model-intrinsic biases that fairness-aware fine-tuning can correct, rather than by recording conditions, dataset artifacts, or unmeasured linguistic factors.

What would settle it

Applying SamaVaani fine-tuning to the held-out psychiatric interview test set and finding neither an overall accuracy gain nor a reduction in gender or role performance gaps would falsify the central claim.

Figures

Figures reproduced from arXiv: 2606.26901 by Abhishek Manoharan, Animesh Mukherjee, Astut Kurariya, Diptadhi Mukherjee, Koustav Rudra, Lekhansh Shukla, Prabhat Chand, Prakrithi Shivaprakash, Pratima Murthy, Subham Kumar.

Figure 1
Figure 1. Figure 1: Architecture for proposed fine-tuning pipeline illustrating LoRA, contrastive and CTC head. (→) indicates fine-tuning flow while (99K) shows inference on test set. as follows. FTStd.: This refers to the standard LoRA fine￾tuning using part of our dataset for training. FTPS: Here we double the dataset by augment￾ing the pitch of the audio files. The hypothe￾sis is that this augmentation of synthetic data wo… view at source ↗
Figure 2
Figure 2. Figure 2: Pitch comparison based on the role of the speaker, i.e, doctor or patient [PITH_FULL_IMAGE:figures/full_fig_p013_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Voice quality comparison based on the role of the speaker, i.e, doctor or patient [PITH_FULL_IMAGE:figures/full_fig_p013_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Pitch comparison based on the gender of the speaker, i.e, male or female [PITH_FULL_IMAGE:figures/full_fig_p013_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Voice quality comparison based on the gen￾der of the speaker, i.e, male or female [PITH_FULL_IMAGE:figures/full_fig_p013_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Speech intelligibility report of the speech data used in this study [PITH_FULL_IMAGE:figures/full_fig_p014_6.png] view at source ↗
read the original abstract

Automatic Speech Recognition (ASR) is increasingly used to document clinical encounters, yet its reliability in multilingual and demographically diverse Indian healthcare context remains largely unknown. In this study, we first conduct the systematic audit of ASR performance on real-world psychiatric interview data spanning Kannada, Hindi and Indian English, comparing eight state-of-the-art models including IndicWhisper, WhisperLargeV3, Sarvam, GoogleS2T, Gemma3n, OmniLingual, Vaani, and Gemini. Our results reveal substantial variability across models and languages, with some systems performing competitively in Indian English but failing in regional speech. We further fine-tune two of the best performing opensource models, i.e., Gemma3n and OmniLingual, using various methods. With this, we uncover systematic performance gaps tied to speaker role and gender, raising concerns about equitable deployment in clinical settings, which are further mitigated by fairness-aware fine-tuning. To this end, we propose SamaVaani, a unified debiasing technique that simultaneously improves ASR performance and improves fairness across demographic groups.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper audits eight ASR models (including IndicWhisper, WhisperLargeV3, Sarvam, GoogleS2T, Gemma3n, OmniLingual, Vaani, and Gemini) on real-world psychiatric interview data in Kannada, Hindi, and Indian English, identifies substantial performance variability and systematic gaps tied to speaker role and gender, and proposes SamaVaani as a unified fairness-aware fine-tuning technique applied to Gemma3n and OmniLingual that simultaneously boosts overall ASR performance and reduces demographic disparities.

Significance. If the empirical results hold under proper controls, the work would be significant for clinical ASR deployment in multilingual Indian healthcare, where equitable performance across languages and speaker demographics is critical for reliable documentation of psychiatric encounters.

major comments (2)
  1. [Methods (fine-tuning and evaluation procedure)] The experimental design for both the audit and the SamaVaani fine-tuning uses the same real-world psychiatric interview corpus without explicit held-out sets or controls differing in recording equipment, clinical sub-domain, or unmeasured linguistic factors. This directly undermines the central claim that performance gaps are primarily model-intrinsic biases amenable to the proposed debiasing, rather than dataset or recording artifacts.
  2. [Results and Experiments] No quantitative results, error bars, dataset sizes, fine-tuning hyperparameters, or validation procedures (e.g., train/test splits, statistical significance tests) are provided, making it impossible to assess whether the reported improvements in performance and fairness are supported by evidence or reproducible.
minor comments (1)
  1. The abstract lists eight models but does not specify which two were selected for fine-tuning beyond naming Gemma3n and OmniLingual; clarify selection criteria.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive comments, which identify key areas where the manuscript requires greater rigor. We respond to each major comment below and commit to revisions that will strengthen the experimental description and evidence presentation.

read point-by-point responses

  1. Referee: [Methods (fine-tuning and evaluation procedure)] The experimental design for both the audit and the SamaVaani fine-tuning uses the same real-world psychiatric interview corpus without explicit held-out sets or controls differing in recording equipment, clinical sub-domain, or unmeasured linguistic factors. This directly undermines the central claim that performance gaps are primarily model-intrinsic biases amenable to the proposed debiasing, rather than dataset or recording artifacts.

    Authors: We agree that the current manuscript lacks explicit detail on held-out sets and controls for recording equipment or other factors, which weakens the ability to isolate model-intrinsic effects. In revision we will expand the Methods section to specify the train/validation/test splits (including ratios and any demographic stratification), describe controls or post-hoc analyses for recording and sub-domain factors where possible, and discuss limitations regarding unmeasured linguistic variables. These changes will clarify the scope of claims about SamaVaani while acknowledging potential dataset influences. revision: yes

  2. Referee: [Results and Experiments] No quantitative results, error bars, dataset sizes, fine-tuning hyperparameters, or validation procedures (e.g., train/test splits, statistical significance tests) are provided, making it impossible to assess whether the reported improvements in performance and fairness are supported by evidence or reproducible.

    Authors: We acknowledge that these quantitative elements were omitted from the submitted version. The revised manuscript will add a dedicated Results section containing dataset sizes (audio hours and utterance counts per language and demographic), all WER and fairness metrics with error bars or confidence intervals, fine-tuning hyperparameters, explicit train/test split details, and statistical significance tests (e.g., paired tests with p-values). This will enable assessment of reproducibility and the strength of the reported gains. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical audit and fine-tuning study

full rationale

The paper reports an empirical audit of eight ASR models on a real-world psychiatric interview corpus in Kannada/Hindi/Indian English, followed by fine-tuning of two models with fairness-aware methods and evaluation of SamaVaani. No equations, derivations, or parameter-fitting steps are described that reduce any claim to its own inputs by construction. No self-citation chains or uniqueness theorems are invoked as load-bearing premises. The central results (performance gaps and mitigation) are presented as direct experimental measurements on the study data, rendering the work self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review yields no identifiable free parameters, axioms, or invented entities; all details on modeling choices and data handling are absent.

pith-pipeline@v0.9.1-grok · 5760 in / 1062 out tokens · 39512 ms · 2026-06-26T04:45:45.871478+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

297 extracted references · 208 canonical work pages

  1. [1]

    Eka Care , year =. The

  2. [2]

    Proceedings of the International AAAI Conference on Web and Social Media , author =

    A. Proceedings of the International AAAI Conference on Web and Social Media , author =. 2024 , pages =. doi:10.1609/icwsm.v18i1.31390 , abstract =

    work page doi:10.1609/icwsm.v18i1.31390 2024

  3. [3]

    , month = may, year =

    Javed, Tahir and Joshi, Sakshi and Nagarajan, Vignesh and Sundaresan, Sai and Nawale, Janki and Raman, Abhigyan and Bhogale, Kaushal and Kumar, Pratyush and Khapra, Mitesh M. , month = may, year =. Svarah:. doi:10.48550/arXiv.2305.15760 , abstract =

    work page doi:10.48550/arxiv.2305.15760

  4. [4]

    , month = dec, year =

    Javed, Tahir and Bhogale, Kaushal Santosh and Raman, Abhigyan and Kunchukuttan, Anoop and Kumar, Pratyush and Khapra, Mitesh M. , month = dec, year =. doi:10.48550/arXiv.2208.11761 , abstract =

    work page doi:10.48550/arxiv.2208.11761

  5. [5]

    SSM2Mel: State Space Model to Reconstruct Mel Spectrogram from the EEG,

    Barański, Mateusz and Jasiński, Jan and Bartolewska, Julitta and Kacprzak, Stanisław and Witkowski, Marcin and Kowalczyk, Konrad , month = apr, year =. Investigation of. doi:10.1109/ICASSP49660.2025.10890105 , abstract =

    work page doi:10.1109/icassp49660.2025.10890105 2025

  6. [6]

    and Vaijayanthi, C

    Javed, Tahir and Nawale, Janki Atul and George, Eldho Ittan and Joshi, Sakshi and Bhogale, Kaushal Santosh and Mehendale, Deovrat and Sethi, Ishvinder Virender and Ananthanarayanan, Aparna and Faquih, Hafsah and Palit, Pratiti and Ravishankar, Sneha and Sukumaran, Saranya and Panchagnula, Tripura and Murali, Sunjay and Gandhi, Kunal Sharad and R, Ambujava...

    work page doi:10.48550/arxiv.2403.01926

  7. [7]

    Accented

    Hinsvark, Arthur and Delworth, Natalie and Rio, Miguel Del and McNamara, Quinten and Dong, Joshua and Westerman, Ryan and Huang, Michelle and Palakapilly, Joseph and Drexler, Jennifer and Pirkin, Ilya and Bhandari, Nishchal and Jette, Miguel , month = jun, year =. Accented. doi:10.48550/arXiv.2104.10747 , abstract =

    work page doi:10.48550/arxiv.2104.10747

  8. [8]

    Proceedings of the National Academy of Sciences , author =

    Racial disparities in automated speech recognition , volume =. Proceedings of the National Academy of Sciences , author =. 2020 , note =. doi:10.1073/pnas.1915768117 , abstract =

    work page doi:10.1073/pnas.1915768117 2020

  9. [9]

    Gender and Dialect Bias in Y ou T ube ' s Automatic Captions

    Tatman, Rachael , editor =. Gender and. Proceedings of the. 2017 , pages =. doi:10.18653/v1/W17-1606 , abstract =

    work page doi:10.18653/v1/w17-1606 2017

  10. [10]

    Psychiatry Research , author =

    Moving beyond word error rate to evaluate automatic speech recognition in clinical samples:. Psychiatry Research , author =. 2025 , keywords =. doi:10.1016/j.psychres.2025.116690 , abstract =

    work page doi:10.1016/j.psychres.2025.116690 2025

  11. [11]

    IEEE Access , author =

    Relative. IEEE Access , author =. 2025 , keywords =. doi:10.1109/ACCESS.2025.3585454 , abstract =

    work page doi:10.1109/access.2025.3585454 2025

  12. [12]

    Assessing

    Ferizaj, Drin and Neumann, Susann , editor =. Assessing. Human-. 2024 , keywords =. doi:10.1007/978-3-031-60449-2_2 , abstract =

    work page doi:10.1007/978-3-031-60449-2_2 2024

  13. [13]

    Journal of Clinical Medicine , author =

    Automated. Journal of Clinical Medicine , author =. 2024 , note =. doi:10.3390/jcm13174997 , abstract =

    work page doi:10.3390/jcm13174997 2024

  14. [14]

    Improved estimation of daily blue-sky snow shortwave albedo from MODIS data and reanalysis information , volume=

    What automatic speech recognition can and cannot do for conversational speech transcription , volume =. Research Methods in Applied Linguistics , author =. 2024 , keywords =. doi:10.1016/j.rmal.2024.100163 , abstract =

    work page doi:10.1016/j.rmal.2024.100163 2024

  15. [15]

    Speech-based

    Chen, Yangbin and Xu, Chenyang and Liang, Chunfeng and Tao, Yanbao and Shi, Chuan , month = jun, year =. Speech-based. doi:10.48550/arXiv.2406.03510 , abstract =

    work page doi:10.48550/arxiv.2406.03510

  16. [16]

    Psychiatry Research , author =

    Combining automatic speech recognition with semantic natural language processing in schizophrenia , volume =. Psychiatry Research , author =. 2023 , keywords =. doi:10.1016/j.psychres.2023.115252 , abstract =

    work page doi:10.1016/j.psychres.2023.115252 2023

  17. [17]

    Artificial Intelligence in Medicine , author =

    Automatic documentation of professional health interactions:. Artificial Intelligence in Medicine , author =. 2023 , keywords =. doi:10.1016/j.artmed.2023.102487 , abstract =

    work page doi:10.1016/j.artmed.2023.102487 2023

  18. [18]

    Revista Colombiana de Psiquiatría , author =

    Automatic. Revista Colombiana de Psiquiatría , author =. 2024 , keywords =. doi:10.1016/j.rcp.2023.12.002 , abstract =

    work page doi:10.1016/j.rcp.2023.12.002 2024

  19. [19]

    Procedia Computer Science , author =

    Challenges of. Procedia Computer Science , author =. 2023 , keywords =. doi:10.1016/j.procs.2023.10.101 , abstract =

    work page doi:10.1016/j.procs.2023.10.101 2023

  20. [20]

    BMC Medical Education , author =

    Do clinical interview transcripts generated by speech recognition software improve clinical reasoning performance in mock patient encounters?. BMC Medical Education , author =. 2023 , keywords =. doi:10.1186/s12909-023-04246-9 , abstract =

    work page doi:10.1186/s12909-023-04246-9 2023

  21. [21]

    Rao, Ravindra and Ambekar, Atul and Agarwal, A , year =. Opioid

  22. [22]

    Indian Journal of Psychiatry , author =

    Sociodemographic and clinical profiles of individuals with opioid dependence syndrome presenting to a tertiary care center in. Indian Journal of Psychiatry , author =. 2024 , pmid =. doi:10.4103/indianjpsychiatry.indianjpsychiatry_907_24 , abstract =

    work page doi:10.4103/indianjpsychiatry.indianjpsychiatry_907_24 2024

  23. [23]

    The New England Journal of Medicine , author =

    Management of opioid analgesic overdose , volume =. The New England Journal of Medicine , author =. 2012 , pmid =. doi:10.1056/NEJMra1202561 , language =

    work page doi:10.1056/nejmra1202561 2012

  24. [24]

    doi:10.1111/ajad.12862 , abstract =

    Review article:. doi:10.1111/ajad.12862 , abstract =

    work page doi:10.1111/ajad.12862

  25. [25]

    2018 , pmid =

    Medications for. 2018 , pmid =

    2018

  26. [26]

    The American Journal of Drug and Alcohol Abuse , author =

    A novel non-opioid protocol for medically supervised opioid withdrawal and transition to antagonist treatment , copyright =. The American Journal of Drug and Alcohol Abuse , author =. 2018 , note =

    2018

  27. [27]

    The Cochrane Database of Systematic Reviews , author =

    Alpha2-adrenergic agonists for the management of opioid withdrawal , issn =. The Cochrane Database of Systematic Reviews , author =. 2009 , pmid =. doi:10.1002/14651858.CD002024.pub3 , abstract =

    work page doi:10.1002/14651858.cd002024.pub3 2009

  28. [28]

    Lancet (London, England) , author =

    New directions in the treatment of opioid withdrawal , volume =. Lancet (London, England) , author =. 2020 , pmid =. doi:10.1016/S0140-6736(20)30852-7 , abstract =

    work page doi:10.1016/s0140-6736(20)30852-7 2020

  29. [29]

    The Cochrane Database of Systematic Reviews , author =

    Buprenorphine for managing opioid withdrawal , volume =. The Cochrane Database of Systematic Reviews , author =. 2017 , pmid =. doi:10.1002/14651858.CD002025.pub5 , abstract =

    work page doi:10.1002/14651858.cd002025.pub5 2017

  30. [30]

    Cureus , author =

    Tramadol for the. Cureus , author =. 2020 , note =. doi:10.7759/cureus.9128 , abstract =

    work page doi:10.7759/cureus.9128 2020

  31. [31]

    Methadone at tapered doses for the management of opioid withdrawal -

  32. [32]

    Industrial Psychiatry Journal , author =

    Prevalence and predictors of internet gaming disorder among adolescents in. Industrial Psychiatry Journal , author =. 2025 , pmid =. doi:10.4103/ipj.ipj_61_25 , abstract =

    work page doi:10.4103/ipj.ipj_61_25 2025

  33. [33]

    Opioid withdrawal scales , url =

    Guidelines Development Group , year =. Opioid withdrawal scales , url =. Guidelines for the

  34. [34]

    Advances in Dual Diagnosis , author =

    Time to integrated care and social and clinical determinants: secondary analysis of a multicenter study in persons with dual diagnosis , volume =. Advances in Dual Diagnosis , author =. 2025 , pages =. doi:10.1108/ADD-02-2025-0007 , abstract =

    work page doi:10.1108/add-02-2025-0007 2025

  35. [35]

    , year =

    Ruiz, Pedro and Strain, Eric C. , year =. Lowinson and

  36. [36]

    International Journal of Molecular Sciences , author =

    The. International Journal of Molecular Sciences , author =. 2019 , note =. doi:10.3390/ijms20174302 , abstract =

    work page doi:10.3390/ijms20174302 2019

  37. [37]

    Current anesthesiology reports , author =

    Insights into the. Current anesthesiology reports , author =. 2020 , pmid =. doi:10.1007/s40140-020-00420-7 , abstract =

    work page doi:10.1007/s40140-020-00420-7 2020

  38. [38]

    Annual Review of Neuroscience , author =

    Endogenous. Annual Review of Neuroscience , author =. 2020 , note =. doi:10.1146/annurev-neuro-110719-095912 , abstract =

    work page doi:10.1146/annurev-neuro-110719-095912 2020

  39. [39]

    The Journal of Clinical Investigation , author =

    Opiate addiction and cocaine addiction: underlying molecular neurobiology and genetics , volume =. The Journal of Clinical Investigation , author =. 2012 , pmid =. doi:10.1172/JCI60390 , language =

    work page doi:10.1172/jci60390 2012

  40. [40]

    Current behavioral neuroscience reports , author =

    Current understanding of the neurobiology of opioid use disorder:. Current behavioral neuroscience reports , author =. 2019 , pmid =. doi:10.1007/s40473-019-0170-4 , abstract =

    work page doi:10.1007/s40473-019-0170-4 2019

  41. [41]

    Frontiers in Human Neuroscience , author =

    The. Frontiers in Human Neuroscience , author =. 2021 , note =. doi:10.3389/fnhum.2021.601905 , abstract =

    work page doi:10.3389/fnhum.2021.601905 2021

  42. [42]

    Rao, Ravindra and Agarwal, A and Ambekar, Atul , year =. Opioid

  43. [43]

    The Cochrane Database of Systematic Reviews , author =

    Opioid agonist treatment for pharmaceutical opioid dependent people , issn =. The Cochrane Database of Systematic Reviews , author =. 2016 , pmid =. doi:10.1002/14651858.CD011117.pub2 , abstract =

    work page doi:10.1002/14651858.cd011117.pub2 2016

  44. [44]

    Indian Journal of Critical Care Medicine : Peer-reviewed, Official Publication of Indian Society of Critical Care Medicine , author =

    Poisoning with. Indian Journal of Critical Care Medicine : Peer-reviewed, Official Publication of Indian Society of Critical Care Medicine , author =. 2019 , pmid =. doi:10.5005/jp-journals-10071-23309 , abstract =

    work page doi:10.5005/jp-journals-10071-23309 2019

  45. [45]

    ResearchGate , month = aug, year =

    (. ResearchGate , month = aug, year =

  46. [46]

    International Journal of Nephrology and Renovascular Disease , author =

    Hypomagnesemia: a clinical perspective , volume =. International Journal of Nephrology and Renovascular Disease , author =. 2014 , pmid =. doi:10.2147/IJNRD.S42054 , abstract =

    work page doi:10.2147/ijnrd.s42054 2014

  47. [47]

    Pharmacotherapy , author =

    Management of. Pharmacotherapy , author =. 2016 , pmid =. doi:10.1002/phar.1770 , abstract =

    work page doi:10.1002/phar.1770 2016

  48. [48]

    Journal of Neurology, Neurosurgery, and Psychiatry , author =

    The. Journal of Neurology, Neurosurgery, and Psychiatry , author =. 1989 , pmid =

    1989

  49. [49]

    The New England Journal of Medicine , author =

    Electrolyte. The New England Journal of Medicine , author =. 2017 , pmid =. doi:10.1056/NEJMra1704724 , language =

    work page doi:10.1056/nejmra1704724 2017

  50. [50]

    Journal of Clinical Medicine , author =

    Hyponatremia in. Journal of Clinical Medicine , author =. 2014 , pmid =. doi:10.3390/jcm4010085 , abstract =

    work page doi:10.3390/jcm4010085 2014

  51. [51]

    Journal of trace elements in medicine and biology: organ of the Society for Minerals and Trace Elements (GMS) , author =

    Pathogenetic mechanisms of hypomagnesemia in alcoholic patients , volume =. Journal of trace elements in medicine and biology: organ of the Society for Minerals and Trace Elements (GMS) , author =. 1995 , pmid =. doi:10.1016/S0946-672X(11)80026-X , abstract =

    work page doi:10.1016/s0946-672x(11)80026-x 1995

  52. [52]

    New England Journal of Medicine , author =

    Recognition and. New England Journal of Medicine , author =. 2014 , note =. doi:10.1056/NEJMra1407298 , abstract =

    work page doi:10.1056/nejmra1407298 2014

  53. [53]

    JAMA , author =

    Pharmacological management of alcohol withdrawal. JAMA , author =. 1997 , pmid =. doi:10.1001/jama.278.2.144 , abstract =

    work page doi:10.1001/jama.278.2.144 1997

  54. [54]

    , year =

    International Narcotics Control Board. , year =. Narcotic drugs 2024:

    2024

  55. [55]

    National

    Ministry of Finance, Government of India , year =. National

  56. [56]

    Our joint commitment to effectively addressing and countering the world drug problem , url =

    United Nations General Assembly , month = apr, year =. Our joint commitment to effectively addressing and countering the world drug problem , url =

  57. [57]

    Suggestions for changes in the

    Addiction Psychiatry Society of India , month = oct, year =. Suggestions for changes in the

  58. [58]

    American Journal of Public Health , author =

    Language and. American Journal of Public Health , author =. 2013 , pages =. doi:10.2105/AJPH.2012.301191 , abstract =

    work page doi:10.2105/ajph.2012.301191 2013

  59. [59]

    Ministry of Law. The

  60. [60]

    Statements following the voting on the

    United Nations Office on Drugs. Statements following the voting on the

  61. [61]

    Transnational Institute (TNI)/International Drug Policy Consortium (IDPC) , author =

    Scheduling in the international drug control system , language =. Transnational Institute (TNI)/International Drug Policy Consortium (IDPC) , author =

  62. [62]

    TNI/IDPC Series on Legislative Reform of Drug Policies , author =

    Fifty. TNI/IDPC Series on Legislative Reform of Drug Policies , author =

  63. [63]

    United Nations Office on Drugs. The

  64. [64]

    Vital. MMWR. Morbidity and Mortality Weekly Report , author =. doi:10.15585/mmwr.mm6709e1 , abstract =

    work page doi:10.15585/mmwr.mm6709e1

  65. [65]

    Smoke and

    Ghosh, Amitava , year =. Smoke and

  66. [66]

    , year =

    Harm Reduction International. , year =. The death penalty for drug offences:

  67. [67]

    , year =

    Public Health Agency of Canada. , year =. Supervised consumption sites:

  68. [68]

    , month = oct, year =

    Expert Panel on the Legislative Review of the Cannabis Act. , month = oct, year =. Legislative review of the

  69. [69]

    Funding and delivery of

    Oregon Secretary of State, Audits Division , month = dec, year =. Funding and delivery of

  70. [70]

    As new data dashboard shows overdoses increased in 2023,

    Oregon Health Authority , month = dec, year =. As new data dashboard shows overdoses increased in 2023,

    2023

  71. [71]

    Oregon Legislative Assembly , year =. House

  72. [72]

    Drug decriminalisation in

  73. [73]

    Harm Reduction Journal , author =

    Fifteen years of heroin-assisted treatment in a. Harm Reduction Journal , author =. 2020 , keywords =. doi:10.1186/s12954-020-00412-0 , abstract =

    work page doi:10.1186/s12954-020-00412-0 2020

  74. [74]

    Addiction (Abingdon, England) , author =

    Does heroin-assisted treatment reduce crime?. Addiction (Abingdon, England) , author =. 2022 , pmid =. doi:10.1111/add.15601 , abstract =

    work page doi:10.1111/add.15601 2022

  75. [75]

    The Lancet Regional Health - Europe , author =

    Germany's cannabis act: a catalyst for. The Lancet Regional Health - Europe , author =. 2024 , pmid =. doi:10.1016/j.lanepe.2024.100929 , abstract =

    work page doi:10.1016/j.lanepe.2024.100929 2024

  76. [76]

    Addiction , author =

    Effects of a sustained heroin shortage in three. Addiction , author =. 2005 , note =. doi:10.1111/j.1360-0443.2005.01094.x , abstract =

    work page doi:10.1111/j.1360-0443.2005.01094.x 2005

  77. [77]

    Substance Abuse Treatment, Prevention, and Policy , author =

    The "lessons" of the. Substance Abuse Treatment, Prevention, and Policy , author =. 2006 , pmid =. doi:10.1186/1747-597X-1-11 , abstract =

    work page doi:10.1186/1747-597x-1-11 2006

  78. [78]

    SSRN Electronic Journal , author =

    Going after the. SSRN Electronic Journal , author =. doi:10.2139/ssrn.3010673 , language =

    work page doi:10.2139/ssrn.3010673

  79. [79]

    Drug and Alcohol Dependence , author =

    Mortality in heroin-assisted treatment in. Drug and Alcohol Dependence , author =. 2005 , pmid =. doi:10.1016/j.drugalcdep.2005.01.005 , abstract =

    work page doi:10.1016/j.drugalcdep.2005.01.005 2005

  80. [80]

    National

    Ministry of Finance, Department of Revenue , year =. National

Showing first 80 references.