Annual Me hodological A chi e Resea ch Re iew
h p://am esea ch e iew.com/index.php/Jou nal/abou
Volume 3, Issue 9 (2025)
Online ISSN P in ISSN
3007-3197
3007-3189
h p://am esea ch e iew.com/index.php/Jou nal/abou
Page 451
E alua ing Dep ession, Anxie y, and S ess in College S uden s:
Combining DASS-21, BAI, BDI, and AI o P e en i e Suppo
Yumna Wa si
Dep . o So wa e Enginee ing, FAST-NUCES, Ka achi, Pakis an
[email p o ec ed]om
Dania Zeh a
Dep . o Compu e Science FAST-NUCES, Ka achi, Pakis an
daniazeh [email protected]
Hi a I ikha
Dep . o Compu e Science FAST-NUCES, Ka achi, Pakis an
[email p o ec ed]m
Ms. Aqsa Fayyaz*
Dep . o Science and Humani ies FAST-NUCES, Ka achi, Pakis an. Co esponding
Au ho Email: [email p o ec ed]
Unde g adua e s uden s a e inc easingly conce ning men al heal h challenges like
s ess, anxie y, and dep ession, which equen ly s ay concealed un il hey in ensi y.
The DASS-21 Scale (Dep ession Anxie y S ess Scales— 21 i ems- solely he s ess
ques ion is employed), BDI (Beck Dep ession In en o y) o dep ession ques ions,
and BAI (Beck Anxie y In en o y) o anxie y ques ions a e ecognized ins umen s
u ilized in his in es iga ion o e alua e hese condi ions. Da a was collec ed om
wo s uden ques iona ies—one cen e ed on gene al wellness and he o he
ega ding dep ession. We analyzed he in o ma ion o ca ego ize sco es based on
in ensi y and examine pa e ns in signs like exhaus ion and hopelessness ollowing
i s handling h ough con en ional echniques such as ackling absen da a.
Subsequen ly, he s udy e alua ed a i icial in elligence h ough he u iliza ion o
h ee machine lea ning models o iden i y hei e icacy in iden i ying hese
condi ions. Relying on ou indings, a Random Fo es model showed ou s anding
e ec i eness in e alua ing deg ees o s ess, anxie y, and dep ession, a aining an
accu acy le el o up o 98.8 demons a es ha me ging con empo a y. Al wi h
classical psychological e alua ions can esul in enhanced, mo e an icipa o y
assis ance o s uden psychological well-being
Key Te ms: College s uden s, dep ession, anxie y, ension, DASS-21, BDI, BAI,
ABSTRACT
Annual Me hodological A chi e Resea ch Re iew
h p://am esea ch e iew.com/index.php/Jou nal/abou
Volume 3, Issue 9 (2025)
Online ISSN P in ISSN
3007-3197
3007-3189
h p://am esea ch e iew.com/index.php/Jou nal/abou
Page 452
machine lea ning, psychological well-being,
Index Te ms: Unde g adua e s uden s, dep ession, anxie y, s ess, DASS-21,
BDI, BAI, a i icial in elligence, men al heal h assessmen , machine lea ning,
Random Fo es
In oduc ion
Men al heal h issues such as s ess, anxie y, and dep ession a e p e alen among
uni e si y s uden s and a e p o oundly in e linked, leading o in ica e issues ha
could wo sen o e ime [2]. Dep ession, cha ac e ized by pe sis en so ow and a
lack o mo i a ion, nega i ely impac s a s uden ’s educa ional success and gene al
heal h (Beck, S ee , and B own, 1396. Heigh ened anxie y and physical es lessness
in- dica e eelings o anxie y, which equen ly occu oge he wi h dep ession and
can escala e he issue (Julian, 2011). S uden s encoun e ing p essu e om socie al
and educa ional s anda ds may lead o emo ional exhaus ion and g ow mo e
suscep ible o a ious psychological conce ns [7]. These p oblems equen ly
in ensi y one ano he , anxie y om wo k can ini ia e o exace ba e anxie y and
dep ession. Academic li e is equen ly linked o he beginnings o his s ess, such as
challenging assignmen s, a cu h oa a ambiance, and s ess ela ed o exams [5]. A
s uden ’s backg ound—like hei en i onmen , gende , and social class — impac s
hei deg ee o suscep ibili y o hese di icul ies. Resea ch indica es ha emale
lea ne s om low-income amilies equen ly encoun e ele a ed le els o anxie y ,
s ess and dep ession [1], [4] ,because o he complexi y o hese issues, inno a i e
app oaches a e equi ed o deli e excellen ca e. In elligen au oma ion (1A)
p o ides a ou e o ac i e o e sigh o men al wellness. A i icial in elligence
acili a es immedia e obse a ion o s uden engagemen , wellness, ea ly de ec ion o
isk ac o s, and ailo ed assis ance. Al sys ems ha e he capabili y o examine da a
o iden i y sub le indica ions o anxie y, o en e en be o e s uden s iden i ying hese
signs [8]. This allows o as e suppo and lessons he bu den on collages,
counseling se ices, and p o ides s uden s wi h esou ces o sel -conce n.
Hypo heses
Main Hypo hesis (HM): P ima y Hypo hesis (PH): Ma- chine lea ning algo i hms
a e capable o accu a ely gauge he se e i y o dep ession, anxie y, and s ess among
college s u- den s acco ding o hei sel - epo ed eedback om li es yle and
Annual Me hodological A chi e Resea ch Re iew
h p://am esea ch e iew.com/index.php/Jou nal/abou
Volume 3, Issue 9 (2025)
Online ISSN P in ISSN
3007-3197
3007-3189
h p://am esea ch e iew.com/index.php/Jou nal/abou
Page 453
psychological su eys.
Seconda y Hypo heses
•H1:S uden s wi h a eco ded his o y o domes ic iolence lence is an icipa ed
o demons a e conside ably g ea e a e age esul s in dep ession, anxie y, and
s ess le els.
•H2:Assigni ican inclina ion owa ds ea ing junk ood is posi i ely associa ed
wi h inc eased sco es on dep ession, anxie y, and s ess indica o s o men al well-
being.
•H3:The ield o s udy a s uden chooses g ea ly signi i- can ly a ec s hei
o e all men al well-being
Li e a u e Re iew
These elemen s g ea ly impac s uden s’ li es and educa ional successes, and hey
a e a ely ega ded on hei own [2]. Schola ly p essu e sou ces a e an
acknowledged ac o esea ch sugges s a no able connec ion be ela ionship be ween
wo kload and anxie y as well as dep ession symp oms [11].
Resea ch consis en ly shows ha s uden s om lowe -income backg ounds
encoun e addi ional challenges ha in luence hei men al well-being [9], while
emale s uden s indica e g ea e le els o s ess, dep ession and anxie y. [15].
Su eys and adi ional assessmen s a e use ul, bu hey ely on s uden s
ac i ely seeking assis ance [12]. A i icial In elligence has become a signi ican ool
o ackle his p ob- lem. By examining digi al da a, me hods such as machine
lea ning is capable o iden i ying ea ly indica o s, and Al d i en applica ions can
p o ide ailo ed in e en ions, im- p o ing a ailabili y o assis ance and dec easing
he s igma linked h ough looking o assis ance [13], [18]. None heless, he e a e
cons ain s linked o Al, encompassing conce ns ega ding da a p i acy and bias
wi hin algo i hms, highligh ing he necessi y o AI o.
Me hodology
We employed wo sepa a e da ase s o his esea ch o showcase an ex ensi e
analysis o s uden s’ men al heal h. The i s da ase included an in-dep h
psychological ques ion- nai e ha included s ess, anxie y and dep ession,
highligh ing a sample size consis ing o 2028 pa icipan s. A supplemen a y da ase
(N=328) ha speci ically aimed a dep ession was included o p omo e a mo e
Annual Me hodological A chi e Resea ch Re iew
h p://am esea ch e iew.com/index.php/Jou nal/abou
Volume 3, Issue 9 (2025)
Online ISSN P in ISSN
3007-3197
3007-3189
h p://am esea ch e iew.com/index.php/Jou nal/abou
Page 454
concen a ed examina ion. Fu he mo e, o in es iga e he impac s o speci ic
li es yle ac o s, his su ey collec ed da a ega ding demog aphics, e e yday
ac i i ies like social media engagemen , backg ound in o ma ion, collage’ load and a
s aigh o wa d ‘men al heal h assessmen ’ ou come, ca ego izing s uden s in o
‘no mal’, ‘mild’, ‘mode a e’, ‘se e e’ and ‘ex eme’ ca ego ies. ca ego ies.
Collec ion and O ganiza ion o Da a
The da a was me iculously p ep ocessed o gua an ee quali y. We employed he
mos common esponse (mode) o ca ego ical. inqui ies and he median igu e o
nume ical-scale inqui ies 1o ill in he blanks. Fo he e alua ion o quan i a i e
measu es, w i en Responses such as ”A ew days” we e changed in o nume ical
alues.
Explo a o y Da a Analysis
The su ey ques ions we e ca ego ized in o h ee main g oups: anxie y, dep ession,
and s ess. A o al sco e was as- signed o each ca ego y, wi h a maximum o 30,
which we hen classi ied in o i e le els o se e i y: No mal, Mild, Mode a e, Se e e,
and Ex eme. To iden i y mo e p ecise ends, we addi ionally examined speci ic
symp oms like a igue and sad- ness. We examined he associa ions among a ious
pa ame e s h ough Pea son’s co ela ion due o he i egula dis ibu ion o he
da a and including he ela ionship be ween li es yle ac o s like junk ood p e e ence
and men al heal h sco es
Implemen a ion o Machine Lea ning Model
U ilizing machine lea ning, we aimed o assess i su ey e- sponses could o ecas
he se e i y o men al heal h p oblems. Two sepa a e asks o p edic i e modeling
we e conduc ed Ini ially, we employed he comp ehensi e DASS-21 da ase o de
de elop models designed o p edic he se e i y o Dep ession, anxie y and s ess,
30% o his da a collec ion was ese ed o es ing, while he o he 70 % was
u ilized o ain ou algo i hms. In addi ion o assessing ou heo ies conce ning
li es yle, a dis inc machine lea ning model was de eloped. The goal was p edic he
ca ego y o men al heal h assessmen by le e aging demog aphic and beha io al
in o ma ion as inpu pa ame e s. We e alua ed h ee di e en models o each ask:
Random Fo es , Decision T ee, and K-Nea es Neighbo s (KNN).
Ou main measu e o success was hei abili y o accu a ely p edic he in ensi y
Annual Me hodological A chi e Resea ch Re iew
h p://am esea ch e iew.com/index.php/Jou nal/abou
Volume 3, Issue 9 (2025)
Online ISSN P in ISSN
3007-3197
3007-3189
h p://am esea ch e iew.com/index.php/Jou nal/abou
Page 455
le el o he es da a.
Resul s
The analysis p oduced mul iple signi ican esul s, s a ing wi h he e ec i eness o
he o ecas ing models and succeeded by an examina ion o he elemen s a ec ing
men al heal h esul s, as de ailed in ou hypo heses.
A.
Pe o mance o P edic i e Models
To e alua e ou p ima y hypo hesis (H1), we analyzed he abili y o e e y machine
lea ning model o classi y he se e i y o S ess, Anxie y, and Dep ession. The
A bi a y he andom Fo es algo i hm has eliably demons a ed i s s a us as he
bes model, su passing he e ec i eness o bo h he Decision T ee and K Nea es
Neighbo s. The able p esen s he model’s pe o mance on he un es ed da a. Upon
compa ing he models, a dis inc on unne became appa en - he Random Fo es
me hod. I consis en ly ou pe o med o he s. I success ully o ecas ed he in ensi y
o Dep ession and S ess wi h a ema kable 98.8% p ecision. Fo Anxie y, he
accu acy was e y high a 97.6%. Fig. 1 displays he dis ibu ion o DASS-21
se e i y ca ego ies, o Anxie y, Dep ession, and S ess in a g oup o 323
pa icipan s.
Table I: Accu acy (%) o h ee models ac oss di e en condi ions.
Condi ion
K-Nea es Neighbo s
Decision T ee
Random Fo es
Dep ession
90.7%
97.6%
98.8%
Anxie y
91.9%
95.3%
97.6%
S ess
94.2%
97.6%
98.8%
Upon compa ing he models, a dis inc on unne became appa en he Random
Fo es me hod. I consis en ly ou pe - o med o he s compa ed o he o he wo
models.
Annual Me hodological A chi e Resea ch Re iew
h p://am esea ch e iew.com/index.php/Jou nal/abou
Volume 3, Issue 9 (2025)
Online ISSN P in ISSN
3007-3197
3007-3189
h p://am esea ch e iew.com/index.php/Jou nal/abou
Page 456
Fig. 1. DASS-21 se e i y le els o Anxie y, Dep ession, and S ess among 323
pa icipan s.
Figu e. 1 displays he dis ibu ion o DASS-21 se e i y ca e- go ies o Anxie y,
Dep ession, and S ess in a g oup o 323 pa icipan s. The esul s show ha ac oss
all h ee si ua ions, a signi ican numbe o s uden s epo ed symp oms classi ied
as Mild and no mal. This highligh s he widesp ead na u e o hese men al heal h
conce ns as only a limi ed g oup o pa icipan s we e included in he Typical
ca ego y
B.
E alua ion o Seconda y Hypo heses
We analyzed he in o ma ion ela ed o ou seconda y hypo heses. o in es iga e
he ac o s a ec ing men al heal h esul s Family Violence (H2): The esul s
s ongly suppo ou hypo hesis, demons a ing a clea end whe e indi iduals
who possess expe ienced mo e in ense o se e e o ms o domes ic abuse show
signi ican ly highe sco es ac oss all h ee a eas o men al heal h egions.
Impo an ly, he wo g oups ha show he highes a e age sco es a e hose who
epo ed” g oping and assaul om” close amily” and” Indi ec ” iolence, wi h
a e age dep ession sco es o 14.0 in bo h cases. Con e sely, he indi iduals
epo ing” No” expe ience hose expe iencing amily iolence had he lowes
a e age sco es o Dep ession (6.8), Anxie y (3.5), and S ess (1.1). This is ob ious
he dis inc ion highligh s ha he expe ience o domes ic iolence, in any
documen ed way, is a signi ican elemen linked o he heigh ened le els o men al
anguish among he s uden s in his s udy (Fig.2)
Annual Me hodological A chi e Resea ch Re iew
h p://am esea ch e iew.com/index.php/Jou nal/abou
Volume 3, Issue 9 (2025)
Online ISSN P in ISSN
3007-3197
3007-3189
h p://am esea ch e iew.com/index.php/Jou nal/abou
Page 457
Fig. 2. Illus a ing dep ession, anxie y, and s ess sco es ac oss expe ience
wi h amily iolence.
Die a y Choice (H3): The esul s show dis inc pa e ns each sco e.
The Dep ession Sco e consis en ly places he o highes ac oss all ood p e e ence
ca ego ies, peaking o hose who choose “Depends on mood” a e age a ing o 8.5)
and emaining ele a ed o hose who a o “Consis en ly unheal hy snacks. On he
o he hand, he S ess Sco e emains ela i ely low and s able ac oss mos ca ego ies,
a e aging a ound 1.0 and 1.5, wi h a no able inc ease solely in he” bo h” and”
Depends on he mood” segmen s speci ically, he Anxie y Sco e shows a simila ,
despi e being less se e e, i ends owa ds he Dep ession Sco e, a aining i s peak
le el o hose who” Always ea junk ood” (4.8). This isual assessmen e eals ha
e en hough dep ession le els s ay pe sis en ly high, anxie y and s ess le els a y
mo e based on die a y choices, whe e unheal hie o inconsis en ea ing pa ens a e
linked o inc eased sco es (Fig. 3)
Fig. 3. A e age Men al Heal h Sco es by Food P e e ence
Academic Fac o s (H4): I was ound pe cei ed Academic s ess is an essen ial ac o .
Annual Me hodological A chi e Resea ch Re iew
h p://am esea ch e iew.com/index.php/Jou nal/abou
Volume 3, Issue 9 (2025)
Online ISSN P in ISSN
3007-3197
3007-3189
h p://am esea ch e iew.com/index.php/Jou nal/abou
Page 458
S uden s who demons a e indica ed ha hei wo kload was s ess ul (“Yes”)
showed conside ably ele a ed mean sco es o Dep ession, Anxie y, and S ess in
ela ion o indi iduals who did no , suppo ing ou hypo hesis (Fig. 4).
Fig. 4. The co ela ion be ween pe cei ed cou se load and men al heal h
sco es.
C.
Model Ve i ica ion and Da a O ganiza ion
A con usion ma ix was gene a ed o e i y he accu acy o ou Random Fo es
algo i hm. The ma ix con i med he model’s alidi y accu acy, showing an
excep ionally low numbe o misclassi ica ions wi h me ely 4 False Posi i es and no
False Nega i es (Fig.5)
Fig. 5. Con usion ma ix gene a ed by he model.
Mo eo e , a PCA (P incipal Componen Analysis) cha was employed o
demons a e he K-Means clus e ing ha sepa a ed a pa icipan in o ’ ulne able’
Annual Me hodological A chi e Resea ch Re iew
h p://am esea ch e iew.com/index.php/Jou nal/abou
Volume 3, Issue 9 (2025)
Online ISSN P in ISSN
3007-3197
3007-3189
h p://am esea ch e iew.com/index.php/Jou nal/abou
Page 459
and ’non- ulne able’ g oups. I appea s ha you ex go cu o . he clea isual
di e en ia ion o he clus e s in he g aph e i ies.
Discussion
This s udy demons a es e ec i ely ha machine lea ning is capable o can be
e ec i ely u ilized o men al heal h assessmen in s uden s. The imp essi e
accu acy o he Random Fo es model, which is backed by he con usion ma ix,
demons a es ha Al can con ol consis en ly in e p e he complex pa e ns p esen
in sel - epo ed ques ionnai e esul s. Howe e , he signi icance o his s udy
ex ends beyond only he accu acy o he model. By analyzing ou seconda y
hypo heses, we ha e iden i ied se e al key ac o s ha a e s ongly connec ed o he
men al heal h o s uden s. The ind- ings show ha a his o y o domes ic iolence, a
p e e ence o junk ood and pe cei ed academic p essu e a e associa ed. wi h
inc eased nega i e men al heal h sco es p o ide c ucial, use ul obse a ions. This
sugges s ha he mos e ec i e Al C i e ia-based assessmen ools should e alua e
symp om sco es oo as ake in o accoun hese impo an con ex ual ac o s. The
impo an connec ion be ween men al heal h and amily iolence heal h conce ns
highligh s he impo ance o auma-in o med suppo po sys ems wi hin academic
ins i- u ions. Simila ly, he link be ween die a y habi s and men al heal h sugges s
ha uni e si y heal h p og ams could be enhanced by in eg a ing die a y ad ice
wi h adi ional men al wellness assis ance. These indings sugges ha an ex ensi e
s a egy, combining Al-based sc eening wi h a ge ed, p ac ical in e en ions
ep esen he mos p omising pa h o wa d. Acknowledging he limi a ions o his
s udy is essen ial, especially i s eliance on sel - epo ed da a and he ela ional
aspec o he indings. Fu u e esea ch should o con- cen a e on es ablishing causal
links and me ging longi udinal da a o acking s uden men al wellness o e ime.
Limi a ions and Recommenda ions
This s udy has se e al limi a ions ha should be acknowl- edged. Fi s , he sample
size, hough in o ma i e, was ela- i ely small and d awn om a single ins i u ion,
limi ing he gene alizabili y o he indings o b oade popula ions.
Second, he use o sel - epo ed da a in oduces he po en ial o esponse bias,
including o e o unde - epo ing o men al heal h symp oms.
