The 150 most-cited of 13,066 Therapy entities, grouped by the diseases, genes and clinical metrics they appear alongside. No clinical taxonomy was supplied, yet checkpoint inhibitors, platinum chemotherapy and the GLP-1 class each fall out as their own branch.
Therapy is the fifth-largest category in the graph and nobody has looked at it whole. The tree shows which treatment families the corpus actually covers, and how much literature sits behind each one.
Two labels for one therapy end up adjacent here, because they are used in the same papers about the same diseases. The worklist below ranks those pairs so a human can merge or reject them one at a time — clustering proposes, a person decides.
Scoring pairs by the papers they literally share does not work here: only
11% of the 11,175 pairs share even one paper, so nearly every
distance is 1 and the tree collapses into a single 142-member blob with a
handful of singletons. Instead each therapy is profiled by the
other entities it co-occurs with — 11 types, from Disease and Gene to
Clinical Metric — weighted log1p × IDF, L2-normalised, and joined
with ward linkage. Two drugs that never share a paper still land
together when they point at the same disease.
Cluster names are read off their members by hand; the branch structure itself is untouched output. Underlined therapies have a wiki page (101 of 150).
Every pair whose names overlap — one contains the other, or they share a content word or a six-letter stem — ranked by how differently the two are actually used (87 pairs; closest 20 shown). A low distance next to an overlapping name is a merge candidate. A split badge means the two landed in different clusters, which usually argues against merging. Not every row is a duplicate: adjuvant therapy and neoadjuvant therapy rank high and are clinically distinct — that is the kind of false positive this list exists to put in front of a person.
| Distance | Larger label | Smaller label | Name match | |
|---|---|---|---|---|
| 0.449 | robot-assisted radical prostatectomy 27 | radical prostatectomy 21 | contains | |
| 0.581 | Endovascular Thrombectomy 23 | thrombectomy 21 | contains | |
| 0.612 | vaccination 65 | influenza vaccine 41 | vaccin… | |
| 0.623 | vaccination 65 | vaccine 65 | vaccin… | |
| 0.680 | adjuvant therapy 22 | neoadjuvant therapy 19 | contains | |
| 0.726 | vaccination 65 | COVID-19 vaccine 39 | vaccin… | |
| 0.772 | vaccine 65 | influenza vaccine 41 | contains | |
| 0.789 | vaccination 65 | HPV vaccine 40 | vaccin… | |
| 0.792 | anti-PD-1 therapy 79 | anti-PD-L1 61 | anti | |
| 0.808 | neoadjuvant immunochemotherapy 32 | neoadjuvant therapy 19 | neoadjuvant | split |
| 0.829 | monoclonal antibody 59 | neutralizing antibody 29 | antibody | |
| 0.847 | vaccine 65 | COVID-19 vaccine 39 | contains | |
| 0.851 | monoclonal antibody 59 | antibody-drug conjugate 36 | antibody | split |
| 0.857 | chemotherapy 203 | neoadjuvant chemotherapy 22 | contains | |
| 0.857 | Targeted Cancer Therapy 24 | Proteolysis-targeting chimeras 20 | target… | |
| 0.862 | fluorouracil 53 | cisplatin/fluorouracil 33 | contains | |
| 0.864 | chemotherapy 203 | Therapy 47 | contains | split |
| 0.871 | influenza vaccine 41 | COVID-19 vaccine 39 | vaccine | |
| 0.876 | checkpoint inhibitor 282 | immune-checkpoint blockade 35 | checkpoint | |
| 0.883 | platinum-based chemotherapy 22 | platinum 18 | contains | split |
Leaves at the right, root at the left; the further left two branches join, the less alike they are. Bars give each therapy's paper count.