Reference database, calculations and verification record for "Bioelectrical Parameters of Living Systems" and its companion manuscripts: 213 verified sources, and every quoted value checked against its source

Dataset. The verified reference database and calculation script behind the narrative review Bioelectrical Parameters of Living Systems (Rohera, 2026). Three manuscripts by the same author draw on this database: the narrative review named above (concept DOI 10.5281/zenodo.22829923), a companion review of the electrical properties of human tissues and cells, and a companion review of voltage, current and frequency in therapeutic electrical stimulation. Each undertakes, in its Availability-of-data statement, to deposit this database and link it from the manuscript record. From version 1.2 this deposit discharges that undertaking for all three. Version 1.1 discharged it only for the first, and said otherwise. Contents references_verified_v1.2.csv — 210 records: author, year, title, journal, volume, pages, PMID, DOI, study type, species or preparation, key finding, and how each source is used in the manuscripts. codebook.md — what every column means, including the two different verification levels. search_log.md — databases, dates, approach, and the stated limits of the search. worked_example_dosimetry.py — reproduces the idealised calculation in Section 5.2, with the four limitations the manuscript states written into the file. README.md — orientation and citation. A crosswalk, so any citation can be checked The column paper1_ref_number maps each record to its numbered reference in the manuscript. 141 of 210 records carry one. Of the 69 records that carry no Paper 1 reference number, 35 (prefixes K, L and M01-M09) were compiled for the companion manuscripts Papers 2 and 6 and are cited there; they were absent from version 1.1 of this deposit, which is the reason for version 1.2. 34 records are background — read and verified during the work but not cited in the final text; they are kept rather than deleted, because a reference database that shows only what survived into the paper hides the selection. Four manuscript references have no indexed row because they are a book, two IEC standards and a guideline. Verification level — read before relying on a value Bibliographic details were checked against the PubMed or Crossref record for every entry that has one. Five records are indexed in neither and carry no PMID and no DOI: A11 (Hille, Ion Channels of Excitable Membranes, a book), F13 and F14 (IEC 60601-1 and IEC 60601-2-10, standards), I16 (the ICMJE Recommendations) and I17 (the COPE guidelines). For these five, bibliographic details were checked against the publisher's or issuing body's own listing, and no claim of indexed verification is made for them. Content verification is a weaker claim: each quantitative value and each characterisation of a finding was checked at the level of the abstract, and at full text only where the full text was openly accessible. It was not checked against the full text of paywalled sources. The manuscript carries an explicit pre-submission condition that the author read those sources in full, and that condition is not yet discharged. What is new in version 1.3, and why it was added. The data are unchanged: references_verified_v1.2.csv, the codebook, the search log and the worked example are byte-identical to the files deposited in version 1.2, and no record was added, removed or altered. What is added is the verification record — nine files documenting a check of every quoted value in all three manuscripts against the indexed record of the source that reported it, together with the pre-flight script the manuscripts are checked with. Coverage: 43 quantity-bearing sentences drawing on 65 sources in Paper 1, 33 sentences on 26 sources in Paper 2, and 29 sentences on 25 sources in Paper 6; all checked. Not one quoted number was wrong. Nineteen defects were found and every one of them was in the language describing where a number came from, not in the number. It is added because one of the manuscripts already said it was here: Paper 2 §11.10 states that the full record is deposited with the manuscript, and when that sentence was published the pass had been run and written up but existed only in internal programme documents. Publishing this version makes the sentence true, and the gap is recorded rather than closed quietly. Seven entries in the claims register read false-uncorrected; all seven belong to Paper 6 version 1.2, all seven are corrected in a version 1.3 of that manuscript which is deliberately not deposited while a journal is reviewing version 1.2, and the verdict stays as it is because it is true of the artifact a reader can download. verification_README.md explains the method and its two limits in full. The title of this version is extended to name the verification record, because the previous title described only half of what the deposit now contains; the concept DOI is unchanged and the earlier versions keep their own titles. What changed in version 1.2, and a corrected search log Thirty-five records were added: K01-K21 and L01-L05, compiled for the companion review of voltage, current and frequency in therapeutic electrical stimulation, and M01-M09, compiled for the companion review of the electrical properties of human tissues and cells. Version 1.1 was deposited as the reference database behind those manuscripts; for two of them that was not true, because their sources had no rows here. This version makes it true rather than narrowing the claim. No record was removed and no record was altered. The added records have not been through an adversarial omission review of the kind described below, and no equivalent audit has been carried out for them. Records M10 (Burkhardt et al., DepressionDC, Lancet 2023) and M11 (Stupp et al., EF-11, Eur J Cancer 2012) were added in manuscript version 1.2. Both are trials whose results are unfavourable to electrical stimulation, and both had been omitted. An internal adversarial review found seven such omissions, all running in the same direction — toward making the author's commercial field look better founded and safer than its sources say. Version 1.0 of the search log claimed that "null and contrary findings were sought deliberately for each electrotherapy area". That claim did not hold, and the log now says so in place of the claim, rather than being quietly amended. Competing interests The compiler is the founder, Chief Executive Officer and Chief Technology Officer of Rohera Healthcare & Technology Pvt. Ltd., Pune, India, which develops a bioelectronic device, and is the named inventor and holder of granted patents relating to it. This work received no external funding; it was self-funded, which is declared as a competing interest rather than as the absence of one: no external sponsor, grant reviewer or independent body reviewed, constrained or verified it. No data from the device appear in this database, and no claim about the device is made. Use of AI-assisted tools Searching, bibliographic verification against indexed records, and compilation were carried out with the assistance of a large language model (Claude, Anthropic) under the author's direction. The author takes full responsibility. No AI tool is an author. Version 1.4 (29 September 2026). Three records are added and the database now holds 213. The added rows are R211, R212 and R213: RATES, the Delphi consensus reporting checklist for transcranial electrical stimulation (PMID 41044265); REPORT-SCS, the minimum reporting standards for spinal cord stimulation in spinal cord injury (PMID 38271712); and the international minimum reporting standards for transcutaneous vagus nerve stimulation (PMID 33854421). They are mapped to references 147, 148 and 149 of the framework manuscript, so 144 of the 213 records now carry a paper1_ref_number and 69 do not - the same 34 background records and 35 companion-manuscript records as before. Bibliographic details for all three were checked against their PubMed records on the day of deposit. For RATES and REPORT-SCS the key_finding column is explicit that verification is at ABSTRACT LEVEL and that the item lists themselves were not obtained, both being paywalled; the vagus nerve standards are open access. The three were added because the framework manuscript's prior-art paragraph had omitted them, which was a defect in that manuscript rather than a matter of space; the manuscript version that cites them is 1.3.0 (DOI 10.5281/zenodo.23043567), deposited in the same sitting as this database version. references_verified_v1.2.csv is retained alongside the new file rather than removed, because the earlier manuscript versions that cite this deposit were checked against it.