SUPPLEMENTARY MATERIALS

Quantum Machine Learning: An Experimental Survey
Erasmo Rivas, James Resto, Luis Rodriguez, Khalil Dajani and Nabeel Alzahrani
School of Computer Science and Engineering, California State University, San Bernardino

The PRISMA 2020 checklist, the evidence-register summary and the executed search
queries appear in the manuscript as Appendices A to C and are not repeated here.

--------------------------------------------------------------------------------
STUDY IDENTIFIER
--------------------------------------------------------------------------------

Every file identifies a study by its reference number in the manuscript. That
number is the single stable identifier across Files S1, S2 and S3.

--------------------------------------------------------------------------------
CONTENTS
--------------------------------------------------------------------------------

File S1   QAIS_Supplementary_File_S1_Criterion_Level_Appraisal_Data.xlsx
          The evidence-appraisal data.
            Read me              scoring rules and known limitations
            Evidence register    all 107 included studies: capability, evidence
                                 band, real-hardware status and the reference
            Criterion scores     anchor levels 0-4 on all seven criteria for the
                                 105 studies with retained vectors, the weighted
                                 total computed from them, the band, and the
                                 three weight-sensitivity totals of Table 6
            Duplicate appraisal  both reviewers' complete score vectors for the
                                 20 double-appraised studies

File S2   QAIS_Supplementary_File_S2_Study_Extraction_Record.xlsx
          The study-level extraction record: 37 of the 46 recorded fields for
          each of the 107 included studies. The Read me sheet lists every field
          included and every field withheld, with the reason for each.

File S3   QAIS_Supplementary_File_S3_Analysis_Notebook.ipynb
          QAIS_Supplementary_File_S3_Analysis_Notebook.html
          A notebook that reads Files S1 and S2 and recomputes every aggregate
          figure the manuscript reports: Table 2, Figure 4, Table 6 and the
          inter-rater statistics of Section 2. Weighted totals are recomputed
          from the anchor levels rather than read from File S1. Each check
          prints the computed value beside the published value and asserts
          that they agree. The .html file is a static copy of the executed
          notebook, readable in any web browser with nothing installed.

requirements.txt   the Python packages and versions File S3 was verified with

--------------------------------------------------------------------------------
RUNNING FILE S3
--------------------------------------------------------------------------------

Place the .ipynb file in the same folder as Files S1 and S2, install the
packages in requirements.txt, and run all cells. All paths are relative; the
notebook reads only Files S1 and S2, never modifies them, and writes two PNG
figures to the same folder.

--------------------------------------------------------------------------------
NOTES ON THE DATA
--------------------------------------------------------------------------------

1. 105 of 107. Two studies were screened, verified and fully extracted, and
   their evidence bands were recorded, but their criterion-level vectors were
   not retained in the appraisal workbook. The project records do not document
   why. They are excluded from every criterion-level computation and included
   in every band- and hardware-level count, with their bands carried as
   recorded. File S1 identifies both.

2. Weighted totals. Every weighted total in File S1 is the weighted sum of the
   seven anchor levels recorded beside it, and every band is the band of that
   total. File S3 recomputes the totals from the anchor levels independently
   and checks that they agree.

3. Provenance. File S2 contains only rows verified in the project provenance
   audit of 31 July 2026. Rows that audit retired, replaced or withdrawn are
   excluded, as are internal identifiers, reviewer information, workflow fields
   and private working notes.

4. Coverage. Study 77 is cited in the manuscript but has no extraction row in
   the project workbook; its row in File S2 carries the citation only and is
   marked accordingly.

5. Source location. The page, section or table of each source report from which
   a value was taken was not recorded during extraction and is not
   reconstructed.

6. Evidence bands appear only in File S1, so Files S1 and S2 cannot disagree
   about them. Capability maturity levels are given in Table 5 of the
   manuscript.
